UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Phenological, growth and reproductive responses to climate variability and experimental warming in eight… Clark, Karin M. 2004

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

Item Metadata

Download

Media
831-ubc_2004-0411.pdf [ 6.16MB ]
Metadata
JSON: 831-1.0091621.json
JSON-LD: 831-1.0091621-ld.json
RDF/XML (Pretty): 831-1.0091621-rdf.xml
RDF/JSON: 831-1.0091621-rdf.json
Turtle: 831-1.0091621-turtle.txt
N-Triples: 831-1.0091621-rdf-ntriples.txt
Original Record: 831-1.0091621-source.json
Full Text
831-1.0091621-fulltext.txt
Citation
831-1.0091621.ris

Full Text

PHENOLOGICAL, G R O W T H AND REPRODUCTIVE R E S P O N S E S TO CLIMATIC VARIABILITY AND EXPERIMENTAL WARMING IN EIGHT ARCTIC PLANT S P E C I E S by KARIN M. C L A R K BSc. Honours, University of Calgary, 1997 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE D E G R E E OF MASTER OF S C I E N C E in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Geography) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August 2004 © Karin M. Clark, 2004 IUBCL T H E U N I V E R S I T Y OF BRITISH C O L U M B I A F A C U L T Y OF G R A D U A T E STUDIES Library Authorization In presenting this thesis in partial fulfillment 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. Name of Author (please print) Date (dd/mrrf/yyyy) ' TITLEOFTHESIS: ^ Y * J C O \ O $ < & \ , ^ r o c o V , r*rA ^ c r r x ^ u r ^ \ » r fe»soahsr.c E^glrsV- T V ^ < " ^ f o o V O D e g r e e : l y ^ W dr\' <CA*>^cr. Year: Department of The University of British Colum Vancouver, BC Canada Col bia ' O grad'.ubc.ca/torms/?formlD=THS page 1 ol 1 last updated: 20-Jul-04 A B S T R A C T Phenology, growth and reproductive measures were taken on a suite of eight low arctic plant species, Oxytropis nigrescens (Pall.) Fischer, Ledum decumbens (Ait.) Lodd., Vaccinium Vitis-idaea L. var. Minus Lodd., Betula glandulosa Michx., Salix L. spp., Saxifraga tricuspidata Rottb., Eriophorum vaginatum L. and Carex aquatilis Wahlenb. var. stans (Drej.) Boott, over a period of six years (1997-2002) at the Tundra Ecosystem Research Station, Daring Lake, Northwest Territories, Canada. These data were related to climatic conditions and measures of previous season's allocation using Canonical Correlation Analysis (CANCOR). Temperature and snowmelt date were the climatic variables most correlated with spring phenology; whereas, temperature, rainfall and snowmelt are most correlated with growth and reproduction. The spring phenology of vernal flowering plants was less influenced by snowmelt than that of aestival flowering plants. Most species showed decreased time to flower with warmer temperatures yet snowmelt and precipitation patterns modified responses. Climatic relationships to growth and reproductive responses were species specific. Two of the study species, B. glandulosa, and C. aquatilis var. stans, showed negative correlations between the previous and current year's allocation to growth and reproduction. Correlations between previous season's reproductive investment and spring phenology were apparent in four of the study species, O. nigrescens, Salix spp., E. vaginatum and S. tricuspidata, A subset of three of the species, L. decumbens, V. Vitis-idaea and E. vaginatum was experimentally warmed using open-topped chambers for two years (2001-2002). Phenological, growth and reproductive responses were compared to ii unwarmed plots. Under warmer temperatures spring phenology was advanced such that the time from snowmelt to flowering contracted approximately five days for all species. However, timing of the end of the flowering phase remained unchanged. Growth of the two evergreen shrubs increased with warming while that of the sedge was decreased. Reproductive effort and success were essentially unchanged except for V. Vitis-idaea where number of flowers was reduced in the first year of warming. In general, the phenological results of the experimental manipulation confirmed those of the observational data; prefloration interval was shortened with increased temperature. The experimental and observational methods of study proved to be complementary approaches to understanding the complex responses of plants to a changing climate. T A B L E OF C O N T E N T S A B S T R A C T ii TABLE OF C O N T E N T S iv LIST OF T A B L E S vi LIST OF F IGURES viii A C K N O W L E D G E M E N T S ix C H A P T E R I - INTRODUCTION 1 1.1 Phenological Relationships to Climate in Arctic Plants 2 1.2 The Importance of Life History, Growth Form and Genetics on Arctic Plant Phenology 4 1.3 Implications of Changing Phenology on Allocation to Growth and Reproduction 7 1.4 Proposed Research Questions 10 C H A P T E R II - CLIMATIC AND PREVIOUS INVESTMENT RELATIONSHIPS TO VARIABILITY IN PLANT P H E N O L O G Y , G R O W T H AND REPRODUCTION 12 2.1 Introduction 12 2.2 Methods 12 2.2.1 Study Site 13 2.2.2 Study Design 14 2.2.3 Data Collection 17 2.2.4 Data Analysis 19 2.3 Results 22 2.4 Discussion 51 iv 2.4 Conclusions 56 C H A P T E R III - EXPERIMENTAL WARMING 59 3.1 Introduction 59 3.2 Methods 60 3.2.1 Data analysis 61 3.3 Results -62 3.4 Discussion 73 3.5 Conclusions 77 C H A P T E R IV - CONCLUSIONS AND RECOMMENDATIONS FOR F U R T H E R R E S E A R C H 79 4.1 Introduction 79 4.2 Synthesis 79 4.3 Recommendations for further work 81 R E F E R E N C E S 82 LIST OF T A B L E S Table 1 - Phenological and quantitative measures for the eight species 18 Table 2 - Distribution attributes and associated transformation 20 Table 3 - Summary of climatic conditions at Daring Lake, NT 23 Table 4 - Variation in flowering, growth and reproduction 25 Table 5 - Output from Canonical Correlation Analysis of data from Oxytropis nigrescens 29 Table 6 - Output from Canonical Correlation Analysis of data from Ledum decumbens 31 Table 7 - Output from Canonical Correlation Analysis of data from Vaccinium Vitis-idaea 32 Table 8 - Output from Canonical Correlation Analysis of data from Betula glandulosa 34 Table 9 - Output from Canonical Correlation Analysis of data from Betula glandulosa (male phenology and catkins) . 3 5 Table 10 - Output from Canonical Correlation Analysis of data from Salix spp. (female phenology and catkins) 37 Table 11 - Output from the Canonical Correlation Analysis of data for male Salix spp. 39 Table 12 - Output from the Canonical Correlation Analysis of data for Sax'rfraga tricuspidata 40 Table 13 - Output from the Canonical Correlation Analysis of data from Eriophorum vaginatum 42 Table 14 - Output from the Canonical Correlation Analysis of data from Carex aquatilis (female phenology and inflorescences) 43 Table 15 -Output from the Canonical Correlation Analysis of data from Carex aquatilis (male phenology and inflorescences) 45 Table 16 - Simple correlations between plant response variables and environment and previous investment variables 46 Table 17 - Mean annual time to leaf-out and flowering in eight sub-arctic plant species 47 Table 18 - Mean annual growth and reproductive investment measures in eight sub-arctic plant species 49 Table 19 - Relationships between previous and current growth and reproductive investment in eight sub-arctic plant species as determined through Canonical Correlation Analysis 50 Table 20 - Comparison of full and partial models for temperature x time relationships 63 Table 21 - Difference in overall mean and mid-day maximum air and soil, temperature between OTCs and control plots in lichen heath and tussock tundra habitat 65 Table 22 - Analysis of Variance output from the comparison of air and soil temperatures between control plots and OTCs over two years 66 Table 23 - Output from Analysis of Variance on plant responses in Ledum decumbens from control plots and within Open-topped chambers 67 vi Table 24 - Output from Analysis of Variance on plant responses in Vaccinium Vitis-idaea from control plots and within Open-topped chambers 69 Table 25 - Output from Analysis of Variance on plant responses in Ehophorum vaginatum from control plots and within Open-topped chambers 71 vii LIST OF F IGURES Figure 1 - Approximate location of the Tundra Ecosystem Research Station 14 Figure 2 - Location of species plots 16 Figure 3 - Mean first flower date for the eight study plant species 24 Figure 4 - Mean annual vegetative growth for the eight study plant species. 27 Figure 5 - Mean annual reproductive investment for seven of the study plant species 28 Figure 6 - Ordination biplot for Oxytropis nigrescens 29 Figure 7 - Ordination biplot for Ledum decumbens 31 Figure 8 - Ordination biplot for Vaccinium Vitis-idaea 32 Figure 9 - Ordination biplot for Betula glandulosa (female phenology and catkins. 34 Figure 10 - Ordination biplot for Betula glandulosa (male phenology and catkins) 35 Figure 11 - Ordination biplot displaying axes 1 and 2 for female data of Salix spp. 37 Figure 12 - Ordination biplot displaying axes 2 and 3 for female data of Salix spp. 38 Figure 13-Ord inat ion biplot for male data of Sa//x spp 39 Figure 14 - Ordination biplot for Saxifraga tricuspidata 40 Figure 15 - Ordination biplot Eriophorum vaginatum 42 Figure 16 - Ordination biplot for Carex aquatilis (female phenology and inflorescences) 43 Figure 17 - Ordination biplot for Carex aquatilis (male phenology and inflorescences) 45 Figure 18 - Figure 18 - Mean hourly temperatures in control and OTCs for two years 64 Figure 19 - Flowering phenology of Ledum decumbens 67 Figure 20 - Annual growth increment and reproductive investment in Ledum decumbens 68 Figure 21 - Flowering phenology in Vaccinium Vitis-idaea 69 Figure 22 - Reproductive investment and success (a) and mean annual branch growth (b) in Vaccinium Vitis-idaea 70 Figure 23 - Flowering phenology in Eriophorum vaginatum 72 Figure 24 - Reproductive effort (a) and mean leaf length (b) in Eriophorum vaginatum 72 viii ACKNOWLEDGEMENTS I would like to thank Steven Matthews for introducing me to the vegetation research being conducted at the Tundra Ecosystem Research Station and for allowing continued involvement in this work over the years. Thanks are also due to Dr. Anne Gunn, Government of the Northwest Territories, for initiating the study in 1996 and Dr. Greg Henry for invaluable advice throughout the project. There have been numerous researchers, students and volunteers over the years that have been involved in data collection, input and analysis. Michael Svoboda laid out the snow transect, plant plots and collected the first year of data in 1996. Others that have meticulously collected data since that time have been Mika Sutherland, Marc d'Entremont, April Desjarlais, John Lee, Joachim Obst, John MacKay, Koren Erler, Krista Kagume, Leslie Wakelyn, Chris O'Brien, Lisa Dyer, Mindy Willet, Jennifer Lange, Susan Abernathy, Kelly Bourassa and many others. Appreciation is expressed to all for their dedication to this project. ix C H A P T E R I - INTRODUCTION There is mounting evidence on a global scale of ecological change due to global warming (Walther et al. 2002). Annual global temperature has increased 0.5°C over the last century (Maxwell 1997). Whether this temperature rise can be attributable to increased concentrations of atmospheric greenhouse gases is uncertain, as it remains within the range of known historic variability (Maxwell 1997). However, many predictions of Global Circulation Models (GCM's), based on a doubling of atmospheric CO2 concentrations (1990 levels), are consistent with current weather patterns. In the Western Canadian Arctic, for example, recent analysis of current climate trends showed an increase in both mean July temperature (0.28°C, from 1960 -1990) and annual precipitation (10% above 1950-1980 normals, from 1890 - 1990), both of which are predicted by G C M ' s to increase (Maxwell 1997). We may be entering the long-term trend in global climate warming that was predicted to occur first and be most severe in polar regions (Maxwell 1992). In the western Arctic, spring is predicted to be the period of greatest warming (Maxwell 1997). Further, with the combined effects of increased annual precipitation, either in the form of snow adding to snow-depth or rain contributing to snowmelt, timing of snowmelt will most likely advance (Maxwell 1997). The growing season is dramatically short in the Arctic and, therefore, spring is necessarily a period of rapid change. Plants will be particularly susceptible to changes in snowmelt as the timing of growth initiation (i.e. emergence phenology) is intricately tied to snow-free date (Galen and Stanton 1991, Inouye and McGuire 1991, Kudo 1991, Woodley and Svoboda 1994, Walker etal. 1995, Molau 1996, Inouye etal. 1 2002). With earlier snowmelt due to rising temperatures and altered precipitation patterns, emergence phenology will occur correspondingly earlier. Given this context of climate change in the Arctic and the particular sensitivity of plants to this change, examination of responses and potential impacts is important. 1.1 Phenoloqical Relationships to Climate in Arctic Plants Timing of snowmelt is Undoubtedly the major determinant of emergence phenology in arctic plants. In addition to snowmelt date, however, secondary drivers of arctic plant phenology are climatic variables such as temperature, incoming solar radiation and precipitation (Rathcke and Lacey 1985). All of these variables have been examined fairly extensively on a global basis and, generally, different variables "control" phenology in different ecosystems (Rathcke and Lacey 1985). In the Arctic however, there is varied evidence for each of these climatic drivers of phenology. Temperature effects on phenology comprise most of the research to date. For example, a meta-analysis of multi-site experiments, by Arft et al. (1999), across the circumpolar north showed that spring phenology advanced significantly across species and sites with a 1-2°C increase in temperature. Their study utilized results from previously unpublished data along with species specific studies including Cassiope tetragona (Molau 1997a), Papaver radicatum (Levesque etal. 1997), Dryas octopetala (Welker et al. 1997), Salix rotundifolia and S. herbacea (Jones et al. 1997), Silene acaulis (Alatalo and Totland 1997) and Carex bigelowii (Stenstrom and Jonsdottir 1997) many of which also reported that time to flower (from snowmelt to flowering) contracted with higher temperatures. 2 Plant phenology is thought to respond to temperature through the accumulation of heat such that a phenological event occurs after a threshold amount of heat is reached (often calculated as the sum of degrees above 0°C). For example, leaf bud burst in two deciduous arctic shrubs, Salix pulchra and Betula nana, was reported to be determined primarily by heat accumulation albeit in a non-linear fashion (Pop ef al. 2000). Similarly, accumulated heat largely determined time to flower in the evergreen, dwarf shrub Cassiope tetragona (Molau 1997a). In addition, an inter-site analysis of Canadian Tundra and Taiga Experimental sites, showed that phenology of a number of low-arctic plant species was strongly related to accumulated heat (Bean and Henry 2002). Incoming global solar radiation has also been proposed as an important determinant of emergence phenology in arctic plants (Marsden 1992). Irradiance is near its annual peak in the arctic spring and as snow recedes, the dark ground surface absorbs substantial radiation dramatically warming the soil. It is perhaps this drastic increase in soil temperature that promotes early spring growth in many arctic plants (Molau 1997b). For example, flowering in Papaver radicatum has been correlated with soil surface temperature with albedo of the substrate having a large influence on flowering time (Levesque et al. 1997). Radiation has been weakly correlated with time to flower in Ranunculus nivalis (Molau 1997a). However, time to flower increases with more frequent precipitation events presumably through the action of lowered soil temperature (Molau 1997a). Of particular importance in the phenology of arctic plants is previous years' climatic conditions. Reproductive buds in arctic plants are developed from one to 3 several years before anthesis (Sorensen 1941, Bliss 1971). Buds of early flowering species over-winter with highly differentiated flower primordia allowing rapid expansion and blooming the following spring (Sorensen 1941). Climatic conditions from previous years will then influence the state of development of the over-wintering flower buds and determine in part the amount of time required for blooming to occur. In their analysis of several species across the Canadian Arctic, Bean and Henry (2002) reported that previous season's accumulated heat had a relatively strong effect on the phenology of high arctic species. Clearly, there are a number of climatic variables potentially influencing spring phenology in arctic plants. These variables are not mutually exclusive and it is likely that a combination of factors play a role in determining phenological patterns. Although several studies have examined relationships between one or a few of these climatic variables and phenology, primarily focusing on temperature, most have not considered the set of variables as a whole to determine which exert the most influence. The ambiguity in these relationships for arctic plants warrants further investigation. 1.2 The Importance of Life History, Growth Form and Genetics on Arctic Plant Phenology Just how responsive a plant or species will be to warmer temperatures and earlier snowmelt will depend on a number of factors. Within a plant's particular life history, some phenological stages are more responsive to temperature and snowmelt than others. Studies show that early phenological events are more sensitive to changes in snowmelt date than later phenological events (Johnstone 4 1995). Therefore, species that flower in the spring (i.e. "vernal" species according to Bliss 1956) will be relatively more affected by earlier snowmelt than species flowering later (i.e. "aestival") (Molau 1997b). Flowering time varies in response to spring weather conditions in early flowering species allowing them to effectively prolong growing season length (Fitter et al. 1995, Molau 1997b). These species have been termed pollen-risk strategists, because their reproductive success is compromised in the early spring through climatic events affecting pollination (Molau 1993). Late flowering species are not as responsive to spring weather and are, therefore, protected from extreme conditions that may occasionally occur. However, as a result, these species will be relatively more constrained by growing season length than early flowering species (Molau 1993, 1997b). These species, termed seed-risk strategists, have their reproductive success compromised through the effects of the onset of winter on ability to produce mature, viable seeds (Molau 1993, 1997b). There are other constraints on flowering phenology that complicate the relationship between climate and phenology. For example, the relationship between temperature and flowering date in temperate plants is thought to be fairly linear (i.e. earlier flowering date with increasing temperatures) except at the latitudinal extremes of a plant's distribution (Sparks et al. 2000). These limits are presumably established by endogenous characteristics such as genetics and morphology of the plant. In essence, phenology is only responsive to climate within a particular range of plasticity and responsiveness is limited beyond this. 5 Similarly, Fitter et al. (1995) reported that plant growth form influenced responsiveness to temperature. Chamaephytes, phanerophytes and geophytes had better goodness-of-fit for regressions of first flowering date with mean monthly temperatures than hemicryptophytes and therophytes. Experimental warming of a hemicryptophyte, Saxifraga oppositifolia, showed that flowering phenology was not responsive to increases in temperature (Stenstrom ef al. 1997). Yet, Dormann and Woodin (2002) explained that similar morphologies may not necessarily display similar physiological characteristics and, therefore, responsiveness may not be closely linked to growth form. For example, a perennial geophyte, Ranunculus acris, showed little response to artificial warming (Totland 1999), yet it is a growth form that according to Fitter et al. (1995) should respond well to temperature. Differences in ecotype have substantial influence on controlling growth over the season in Eriophorum vaginatum (Fetcher and Shaver 1990). In their study, plants from high arctic sites were not able to increase growth over a prolonged growing season when transplanted to a low arctic site, whereas, plants from the low arctic site were better able to capitalize on a longer growing season. This suggests that ecotypes from stressed sites are not able to respond as completely to an amelioration of the environment as ecotypes from less stressed sites (Fetcher and Shaver 1990). This implies that responses to potential warming will occur over the long term as either better suited ecotypes migrate into an area or that more responsive individuals are selected from among the current population (Fetcher and Shaver 1990). 6 There are numerous factors other than climatic conditions that can play a role in determining plant phenology such as flowering strategy, morphology and ecotype. Low sensitivity to climate may indicate that these types of constraints are relatively more important than climate in determining phenological patterns. These types of responses would also indicate that fairly long periods of time would be necessary for plant phenology to adjust to rapid climate change. 1.3 Implications of Changing Phenology on Allocation to Growth and Reproduction Changes in phenology have implications for both growth and reproduction in arctic plants. The relationship between phenology and vegetative growth in arctic plants is ambiguous. In addition to low temperatures and nutrient availability, plant production in the Arctic is thought to be largely constrained by time, given the relatively short frost-free period available for growth (Galen and Stanton 1991). Longer growing seasons will likely result in increased vegetative growth. Earlier phenology cued by earlier snowmelt will effectively lengthen growing season, even if the onset of winter occurs at the usual time. However, Walker et al. (1995) found that differences in growth in an alpine tundra species could not be predicted from phenology alone (e.g. timing of growth initiation). Instead, a combination of site specific factors, such as nutrient availability, moisture and/or ecotypic variation, determined growth response. In some species exhibiting periodic growth, for example, earlier emergence phenology is offset by earlier senescence such that effective growing season length is unaltered (Starr et al. 2000). Those species with leaf out phenology closely associated with snowmelt will likely show greater growth 7 responses to growing season length than those species with phenology not closely associated with snowmelt (Galen and Stanton 1995). Reproductive effort and success in tundra plants is also linked to snowmelt, emergence phenology and growing season length, although relationships are not simple. Reproductive effort is the investment in reproductive structures (e.g. flower production) whereas reproductive success is a measure of seed viability or production. Relationships between phenology and these two aspects of reproduction differ. In terms of reproductive effort, in some cases, later snowmelt resulted in the production of fewer flowers, as was reported for 56 alpine species (Kudo 1991). In other instances, fewer flowers are produced when snowmelt is advanced, as in two alpine perennial herbs, Delphinium nelsonii, and Polensonium foliosissimum Gray, (Zimmerman and Gross 1984, Inouye and McGuire 1991). There discrepancies in the literature as to how snowmelt and phenology influence reproductive effort. Perhaps the differences are due to the complicating influence of the previous season's climate on bud production combined with effects of snowpack. The maximum number and the phenological state of buds available are determined by the conditions of the year in which buds were formed. However, this would be mediated by the effects of snowpack on the preservation (through insulation) or destruction (through frost damage) of the buds (Inouye and McGuire 1991). Relationships between phenology and reproductive success appear to be clearer. Most studies on arctic and alpine plants show that later phenology results in reduced reproductive success. For example, Galen and Stanton (1991) reported 8 that Ranunculus adoneus plants that flowered late due to persistence of snow had smaller seeds. Similarly, a study by Sandvik and Totland (2000) also showed that delayed flowering in Saxifraga stellaris L. reduced the number of seeds produced per plant. However, the latter study is complicated by the fact that the plants were artificially warmed with open-topped chambers and, therefore, changes in reproductive success cannot solely be attributed to differences in phenology. Perhaps more important than the separate responses of growth and reproduction to phenology is the pattern of allocation to each. In the few long-term studies that have occurred in the arctic, responses have been complex interactions of allocation patterns over time. In their meta-analysis of tundra plant responses to artificial warming, Arft et al. (1999) reported that although phenology was advanced across the time-scale of the study, growth and reproductive responses varied. Growth increased with warming in the first years of the experiment, whereas responses shifted toward increased reproductive effort in the later years. This suggests that there are primary, short-term responses that differ from secondary, longer-term responses. The lack of reproductive response in the first years of the experiments may be indicative of the lag time associated with the development of reproductive structures in arctic plants. Because flower buds are developed one to several years before they open, reproductive responses to experimental manipulations will not manifest for several years. For example, Sandvik and Totland (2000) found that a positive reproductive response in Saxifraga stellaris to warming was not observed until one year after initiation of the experiment. 9 However, this pattern of allocation response is not consistent in the literature. There appeared not to be a lag in reproductive effort and success in Ranunculus acris with experimental warming (Totland 1999). In this case reproductive effort and success were improved with artificial warming over the first three years of the experiment but were not maintained in the fourth year. Further, Johnstone and Henry (1997) reported annual alternations of growth and reproductive responses of Cassiope tetragona apparently as internal resources were utilized and exhausted. Thus, allocation patterns over time are not easily accounted for. Perhaps it is simply that responses, regardless of their initial nature, change with time indicating complex patterns of resource accumulation, storage and allocation. 1.4 Proposed Research Questions In summary, consistent with predictions of G C M s we are experiencing warmer temperatures and greater precipitation in the Western Canadian Arctic. Warmer springs combined with increased precipitation (in the form of rain) will lead to an earlier onset of the snow-free period. Arctic plants will respond to increased temperature, precipitation and altered timing of snowmelt by adjusting their phenology, growth and reproduction patterns. Responses among species will vary, as will the sensitivity of particular phenological stages within a species. The discussion above highlights several areas in which knowledge is scant or where there is sufficient divergence in the literature to warrant further examination. The following research questions are posed in light of these identified gaps and will be addressed in the following two chapters: 10 1. How variable are phenology, growth and reproduction patterns within and among a set of arctic plant species? 2. Is climate linked to this variability and, if so, which climatic variables are most strongly related? 3. How do arctic plants respond when artificially warmed and is this response consistent with predictions based on the observed climatic relationships? 11 C H A P T E R II - CLIMATIC AND PREVIOUS INVESTMENT RELATIONSHIPS TO VARIABILITY IN PLANT P H E N O L O G Y , G R O W T H AND REPRODUCTION 2.1 Introduction In the preceding chapter I demonstrated that climatic variables most likely act synergistically to influence phenological patterns in arctic plants, and these relationships are not simple. Genetics, including growth form and flowering strategy, have the potential to constrain plant responses to climatic variability. Further, phenological patterns are inextricably linked to growth and reproductive patterns over time and, therefore, changes in one aspect will necessarily lead to changes in the others. Within the context of climate warming in the Arctic, the study of plant responses in these regions and linkages to climate and allocation patterns is timely. The variability in phenology, growth and reproduction in eight arctic plant species and the relationships between this variability and climate was examined and is presented in this chapter. 2.2 Methods The study design was developed in accordance with the protocols of the International Tundra Experiment (ITEX), which provides standard methods for measuring phenology, growth and reproduction in arctic plants (Molau and Edlund 1996) and relating these measurements to climate (Johnstone et al. 1996). The use of these protocols ensures that the methods are valid and have a proven track record for use in arctic studies. More importantly, they ultimately allow the comparison of results from this work to other studies across the ITEX network. This 12 research contributes substantially to the Canadian network of ITEX research (CANTTEX) as there are only approximately half a dozen ongoing projects in the country and only one other study that is situated in the inland regions of the Low Arctic. 2.2.1 Study Site The study was initiated in 1996 at the Tundra Ecosystem Research Station at Daring Lake, Northwest Territories, approximately 300 km north of Yellowknife at 64° 52'N, 111° 35'W in the low arctic tundra (Figure 1). The site is within the Canadian Shield physiographic region and displays features typical of this region (Mathews and Morrow 1985), Granitic outcrops and boulder fields are frequent on the landscape as are small oligotrophic lakes. Glacial features, such as eskers, are also prominent, ranging in size from 50 -100 m wide and 10's of km long. This region is within the zone of continuous permafrost, which, at the height of the growing season, varies from 30 cm to 100 cm from the ground surface depending on substrate and vegetative cover. As a result of permafrost and cool summer temperatures, soils are poorly developed. Vegetation in this region consists mostly of lichen tundra dominated by heaths and dwarf shrubs with sedge meadows in poorly-drained areas (Matthews et al. 2001). Except for rare, very small stands of stunted black spruce, Picea mariana (Mill), trees are absent from the region. The treeline is approximately 75 km to the southwest. 13 Figure 1 - Location of the Tundra Ecosystem Research Station 2.2.2 Study Design The landscape in the study area ranges from dry esker tops to water-logged lowlands and the set of study species represents a diversity of growth forms and flowering strategies adapted to this range of habitats. The eight study species are: Oxytropis nigrescens (Pall.) Fischer (forb), Ledum decumbens (Ait.) Lodd. (evergreen dwarf shrub), Vaccinium Vitis-idaea L. var. minus Lodd.(evergreen dwarf shrub), Betula glandulosa Michx. (deciduous dwarf shrub), Salix spp. L. (deciduous dwarf shrub), Saxifraga tricuspidata Rottb. (evergreen dwarf shrub), Eriophorum vaginatum L. (early flowering sedge), and Carex aquatilis Wahlenb. var. stans (drej. Boott. (late flowering sedge). These plant species characterize the typical vegetation of this particular site and were also selected because they represent important caribou, Rangifer tarandus, forage. 14 In terms of spatial organization, each plant species was examined at a single location along a transect. Seven plots (each containing one study species except one plot which contains two), approximately 10 m X 10 m square, are situated along a gradient extending through habitats ranging from xeric sites, at the top of an esker, down through mesic sites at mid-slope, to hydric sites in a wet lowland (Figure 2). O. nigrescens and S. tricuspidata inhabit xeric sites, V. Vitis-idaea, L. decumbens, B. glandulosa and Salix spp. are situated in mesic sites and E. vaginatum and C. aquatilis are located in hydric sites. The sampling unit varies among the study species, but is consistent with the protocols of the ITEX Manual (Molau and Edlund 1996). In cases where individuals were easy to distinguish and fairly compact in size, the entire plant was sampled (e.g. S. tricuspidata and O. nigrescens). In many arctic plant species, however, specific individuals are difficult to distinguish. For example, C. aquatilis spreads laterally via rhizomes and, therefore, genetic individuals are not obviously discerned aboveground. For this species, one tiller was sampled and spaced sufficiently far from other sampling units (approximately 1 m) to ensure relative confidence that sampling units are separate genetic individuals. E. vaginatum forms a tussock that is assumed to be of similar genetic origin and was, therefore, used as the sampling unit. For the other species, B. glandulosa, Salix spp., V. Vitis- idaea and L. decumbens, the individuals can be quite, therefore a subunit of the individual, a single branch, was sampled. 15 Figure 2 - Location of Species Plots (A - O. nigrescens; B-L. decumbens, V. Vitis-idaea; C-B. glandulosa; D - Salix spp.; E - S. tricuspidata; F - E. vaginatum; G - C. aquatilis) Twenty samples of each species were permanently marked and measured yearly. When a tagged individual died, a new individual was tagged and given a new number. The use of randomly selected individuals followed through time is much more powerful in detecting change than independent samples as they control for individual variability thus reducing variance that might arise due to random sampling every year (Elzinga et al. 1998). They do, however, present unique challenges for data analysis that must be addressed. 1 6 2.2.3 Data Collection Phenological events and quantitative measurements recorded for each species varied somewhat because different species have slightly different physiognomies and phenologies. ITEX protocols were followed when they existed for a particular species. When they did not exist, protocols from other similar species were used and slightly modified to be applicable to that species (Table 1). The phenological events represent the major "achievements" of the plants over the growing season and represent the timing of growth initiation, flowering initiation, flower duration and in some cases (B. glandulosa and Salix spp.) the end of the growing season. Plants were visited daily from end of May through to early August and the date of occurrence of a phenological event was recorded. Dates were recorded in day numbers (consecutively numbered days of the year starting with January 1) for statistical analyses. Counts of reproductive structures were taken as they become apparent (e.g. as plants bloom) and every few days until a maximum number was reached. Quantitative measures of growth were taken annually in mid-August. 17 Table 1 - Phenological and quantitative measures for the eight study species. Phenological events were recorded as the day number that the event took place. Species Phenology Quantitative measures 0. nigrescens P1 - snow-free P2a - first green leaf visible P2b - first flower bud visible P3 - first flower open P4 - first petal drop P5 - last petal drop P6 - first seed shed Q1 - number of buds 0.2 - number of flowers Q3 - number of seed pods Q4 - diameter of plant (cm) L. decumbens P1 - snow-free P2 - flower buds visible P3 - first flower open P4 - first flower shed P5 - last flower shed P6 - first fruit visible Q1 - number of flowering stalks Q2 - number of flowers per stalk Q3 - length of annual growth increment (mm) V. Vitis-idaea P1 - snow-free P2 - flower buds visible P3 - first flower open P4 - first flower shed P5 - last flower shed P6 - first fruit visible Q1 - number of flowers Q2 - number of fruit Q3 - length of annual growth increment (mm) B. glandulosa P1 - snow-free P2 - first leaf bud burst P3 - first catkins visible P4a - first stigmas visible P4b - first pollen shed PS - last green leaf turning rusty P6 - first abscission of leaves P7 - all leaves shed Q1 - number of male catkins Q2 - number of female catkins Q3 - length of longest leaf (mm) Q4 - length of annual growth increment (mm) Salix spp. P1 - snow-free P2 - first leaf bud burst P3a - first stigmas visible P3b - first pollen shed P4 - first yellowing of leaves P5 - last green leaf turning yellow P6 - first abscission of leaves P7 - all leaves shed P8 - onset of seed dispersal Q1 - total number of catkins Q2 - length of longest leaf (mm) Q3 - weight of largest leaf (mg) Q4 - total number of mature female catkins Q5 - catkin length (mm) Q6 - length of annual growth increment (mm) S. tricuspidata P1 - snow-free P2 - first new leaves P3 - flower buds visible P4 - first flower open P5 - first petal shed P6 - last petal shed Q1 - number of flower stalks Q2 - plant diameter (cm) E. vaginatum Pi-snow-free P2 - first new leaf visible P3 - first inflorescence bud P4 - first exposed anthers P5 - first seed shed Q1 - number of flower stalks Q2 - inflorescence shaft length (early and late season) (cm) Q3 - mean length of 10 longest leaves (cm) Q4 - tussock diameter to tips of leaves (cm) C. aquatilis P1 - snow-free P2 - first new leaf visible P3 - first stigmas visible P4 - first exposed anthers ' P5 - first yellowing of leaves P6 - first seed shed Q1 - age class of shoot in flower Q2 - length of flowering stem at full flower (cm) Q3 - length of all green leaves (cm) Climate data were acquired from a remote meteorological station located approximately in the centre of the vegetation transect adjacent to plot E (Figure 2). 18 The station is battery powered and charged by a solar panel and measures the following variables every minute which are averaged to give hourly and daily means. • solar radiation (KW/m 2) 2.2.4 Data Analysis Variability in phenology, growth and reproduction in the study plants was assessed using simple means, standard deviations and coefficients of variation. Canonical Correlation Analysis (CANCOR) was used to assess the climatic relationships to phenology, growth and reproduction, as it is particularly suited to the analysis of data where variables within a set are correlated (Dillon and Goldstein 1984, Tabachnick and Fidell 1996). Because samples were measured repeatedly in time, statistical hypothesis testing would not be valid. As a result, C A N C O R was used to evaluate multivariate correlations descriptively. The multivariate correlations were assessed against simple correlations for verification. It has been reported that C A N C O R analysis is enhanced if distributional assumptions are met (Tabachnick and Fidell 1996), therefore data were checked for univariate normality by assessing skewness and kurtosis values. Based on the shape of the distribution, transformations were done according to recommendations in Tabachnick and Fidell (1996) (Table 2). air temp (°C) relative humidity wind speed (m/s) wind direction (compass bearing) , rain fall (mm) soil temperature soil temperature soil temperature soil temperature soil temperature snow depth (cm) surface (°C) 5 cm (°C) 10 cm (°C) 20 cm (°C) 40 cm (°C) 19 Table 2 - Distribution attributes and associated transformation (from Tabachnick and Fidell 1996.) Distribution Transformation Moderate positive skewness Square root Substantial positive skewness Log(10) —with zero Log(10)(x + C) Severe positive skewness, L-shaped 1 Ix —with zero 1/(x + C) Moderate negative skewness S Q R T ( K - x ) * Substantial negative skewness Log(10)(K-x) * Severe negative skewness, J - shaped 1 /( K - x ) * C = a constant added to each score so that the smallest score is 1 K = constant from which each score is subtracted so that the smallest score is 1, usually equal to largest score + 1 x = each data point •called "reflect", changes a negative skew into a positive one then transforms Because of limited sample sizes, only one phenological event, typically date of first flower opening, was used in the analyses for most species, along with a measure of growth and reproduction. In some cases, as with the deciduous shrubs, date of first leaf emergence was used as a second phenological event. In other cases, there was insufficient data to allow the use of a growth measurement (e.g. L. decumbens and V.Vitis-idaea). Further, there was no indication of reproductive investment for C. aquatilis because we only collected data on flowering plants as opposed to a sample of flowering and non-flowering plants. All analyses included a suite of investment and environmental variables: previous reproductive investment, previous growth investment, snow-free date, mean air temperature, thawing degree days (TDD), rainfall and radiation. Previous year's investment in reproduction and growth was included to examine whether there were any discernable patterns in allocation. Phenological events were standardized for snow-free date (date of event - snow-free date); therefore, prefloration and leaf-out times are the number of days from snow-free date to the date of the phenological event (onset of flowering and green-up, respectively). 20 Floration is the number of days from first flower open to first flower shed. Where a single species had separate male and female flowers or individuals, analyses were done separately as the male and female responses constituted separate samples. Counts of flowers in the spring of one season were used as the reproductive investment of the previous year, because flower primordia are formed in the fall of the year prior to opening. In interpreting the C A N C O R results, only those canonical variates with a canonical correlation greater than 0.6 were interpreted. The canonical cross-loadings (correlation between the variable from one set with the canonical variate of the other set) were interpreted rather than the canonical loadings (correlation of the original variable with its respective canonical variate) or canonical coefficients (analogous to coefficients in multiple regression) as cross-loadings are more conservative, less inflated and not as susceptible to instability (Dillon and Goldstein 1984). Cross-loadings with absolute values above 0.3 were interpreted, while those below this were considered of negligible value (Tabachnick and Fidell 1996). Biplots of covariates were constructed from the results of the C A N C O R procedure. Overall plant responses were represented with a point and environmental variables with arrows. The points and tips of arrows were calculated from the canonical loadings and scaled with the eigenvalue of the corresponding axis as follows: r j s * \ (A s /n) where: r j s = intraset correlation of environmental variable j with axis s A s = eigenvalue of axis s n = number of plants 21 This approach is recommended by ter Braak (1995) and provides a least-squares approximation of the covariances between plant responses and environmental variables. All statistical analyses were conducted with the computer software program S A S version 8. 2.3 Results Mean July temperature at the site is 13.8 °C and the average length of the growing season is 103.8 days (Table 3). Precipitation averages 281.9 millimetres over the summer months (JJA) while average cumulative growing degree days is 980.1. E. vaginatum was the first species to flower (day 154) (Figure 3); C. aquatilis was the last (day 187 for female inflorescences and 192 for male). Using the classification of Bliss (1956), E. vaginatum, Salix spp., 8. glandulosa and O. nigrescens are vernal flowering plants while L decumbens, S. tricuspidata, V.Vitis-idaea and C. aquatilis are aestival flowering plants. Of those plants with both male and female flowers, 8. glandulosa, Salix spp. and C. aquatilis, the male flowers distributed pollen slightly after the female flowers of the same species became receptive. 22 Table 3 - Summary of climatic conditions at Daring Lake, NT based on data obtained from a remote meteorological station at the site. Values are six year means (1997-2002). Radiation July mean global solar radiation 22.0 ± 4.5 Mj/m2/day Maximum photoperiod 21 hours 52 minutes Temperature Air (July mean) 13.8±3.8°C Soil (July mean at 20cm depth) 9.1 ±1 .6 °C Cumulative degree days (snowmelt to end of August) 980.1 ±108.9 Length of growing season (snowmelt to first frost) 103.8 ±13.5 days Precipitation Maximum snow depth 40.5 ± 13.0 cm Rain 281.9 ±159.6 mm Wind July mean 4.0 ± 1.9 m/s Mean date of first flower in all the species was not highly variable, with coefficient of variation (CV) ranging from 2.3 to 5.0% (Table 4). O. nigrescens, B. glandulosa and S. tricuspidata had mean flowering dates that were less variable than the snow-free date for their plot, while the remaining five species had mean flowering dates that were more variable than snow-free date. E. vaginatum had the most variable mean first flower date with a C V of 5.0%; female B. glandulosa catkins had the least variable mean date of first flower with a C V of 2.3%. In contrast, the C V for prefloration intervals was much larger, ranging from 13.6% in V. Vitis-idaea to 158.6% in E. vaginatum. Generally, the vernal flowering species had higher variability in time to flower than the aestival flowering species with the exception of C. aquatilis which had the highest variability in the aestival flowering group and was similar to B. glandulosa the least variable of the vernal group. 23 is » 1 «0 <D <0 X O 2 v. at 2 ® o *• to oo g> E -o ™ -a £ w c n e r a " E a c & E .2 • 3 1=1 Jo IO UJ ? «B ' = CD CO jS TO CO CO "O -T *0 + S s » • .2 i» oo *c •° «fco Lu|°° . •2* • s-. # * * CM c cn c re "5. «o «o 5 2 > , CD S: = H o» . 0) © O r e a> > E £ a „ g ?3 . fig g> a> . « 1.2 o> 3. * " • 0 0 •— 5 «'*••» C0 o n? CD " I f ^ 3 = C N I CD C 0) - CO = CO .5>re .2 U. T> OQ E E £ 2 3 *• CO I I I I o> = •a © "E ~ * £ •o £ 0 1 co © g 2 8 » E g > i i 0) U-3 CO U © 5 « O 0) v. ra o •c *• 5 8 12 « !fc CO 5 T5 © o o t 3 O ° -<o cu o f - I o E » l » o 2 ^ £ £ 8 2 •« * • ? i- « *; O £ m ±, h- CO - •= cn T - co © _ o > re 5 * » o >. c .= to £ .2 > E re ° ~ i» « 8 > o 2 3 $ ffl <0 (0 »#— H "O O .O. X) cn c ^ c CO 3 CO 3 c •-.e cD £ £ « CD _^  8 ^ 3 CO cu t *— r-\ CD CO CD CO 73 I i E>t cu 50 0)>£ IB jo 2 t O CD ^ t O CD o)>H > g 0 £ O > 5 > 5 CD CD LL UJ 73 tu -0 Q -0 Q -a LU E CO D .E >. © ©° 5 -H 76 CO > CD > "CD "CD CO > "CD CO > 0 2 c U—» cn '-4—» CO s c to C 1 00 — CD CD cu CD CD CL> CD CD LL. (0 > CD CO > > CD > CD © > t> £ 3, *> U •= £ 5 a. it-cD c 0^  0^  *- t O r~-C *f ai 10 — C © 0 c © « CM T— atio (stm Meal • atio (stm Meal h- CD -s CO LO -P OO Vari app cb 0 CD LO ai CO Vari app 1^ ai CO CM O CM Vari app 00 LO z sz sz 0 0 c CD CO CD CD .2 x: 0J ^ Jo «1 0^ s? 0^ ^ 0 vP VP 0^  0^  0^  0^  0^  0^  5^  5 0 C- LO CD CD LO CO CM O CD CO CO c > O) 00 0 T- CD O CO LO ai CO LO co T- LO CM CD CO CO CM © iT p^ CM 00 s~ 0 C «- Q. CNJ CM 1^  x— 0 S co CM CM V > © gS co 0 S ~8 0^  -e 0^  0^  0^ 0  p^ CO Vari TO II whe LO CO CD CO LO LO O CO CD 10 LO CO LO 0 Vari TO II whe CM T _ T _ CM CM T _ c S c CO 0^  CO 0s-CD 0 ,~ CM ci to s> •= * 2 >-P 0^ > ^ P^ 0^  0^  0^  0^  n o n O O) O O CM 00 O CN > c •o CO CO CO CO LO CO iation now-! date fie plo 0^ •vP P^ 1- CO W +5 0^  0^  0^  0^  0^  ™ C £ ^ ^ — «- CO 00 LO CO CO h-co °? LO LO CD T— CD O 10 CO co c CO c CO co s « © © O 3 .CO CO © 0 CO £ s 3 O c co ci Q. CO O <S CO .c 5) 5 CO 3 "0 © "5> i< CD l> © c 5 *> CD a. a 00 "5 Uj Cj to O - J <0 10 CN Variability in annual growth ranged from 16.9% in leaf size of 6. glandulosa to 61% in branch growth of Salix spp. Leaf growth in both B. glandulosa and Salix spp. was much less variable than annual branch growth. There appeared to be no trends in annual growth variability among species with similar flowering strategy (Figure 4). Variability in annual reproductive investment was much higher than that of growth and appeared to be larger in vernal species (ranging from 105.7% in female Salix spp. to 249% in female B. glandulosa) as compared to aestival (ranging from 59.6% in L decumbens to 105.5% in S. tricuspidata). Again, the lowest variability in the vernal species was virtually the same as the highest variability in the aestival species (Figure 5). In terms of the climatic relationships to plant responses, C A N C O R analysis of the data for O. nigrescens showed two pairs of canonical variates with canonical correlations greater than 0.6 (Table 5). The first set of canonical variates showed that plant diameter, number of flower buds and prefloration time were positively influenced by rainfall and snow-free date and negatively influenced by TDD and radiation (Figure 6). The second canonical variate showed that plant diameter and prefloration were also negatively influenced by number of flower buds from the previous season. The amount of variance in plant response accounted for by the first environmental canonical variate was 50%, while that of the second was 10%, for a total of 60%. 26 \ •>rum vaginatum fraga tricuspidata -co irex aquatilis •I < < -Ct ; nigrescens -Oxytropit ilix spp. andulosa imbens s-idaea CO <* 3 ^ T 1 ~ E L T, E .3 f 00 1 * 1 o o o in CM O o 2 CD CD E o — - E o o o o to TJ C CO c re CN r c m X E « 2 co Q. O 0) a O) g i n s .2 . \o -e > on a > CD o " CM C •goo.* « « °> o «3 CD ~; C . • !5 N C + ^'5 e E io 1 8 2 o — o CO = £ g " J- CD 2 -° iu § . u CO ® 0) % CD 73 = c1 p CO 43 41 I o o CD a. • co o c °» re «2 S TJ ® 2 re I 2 « § £ . ' 2 •s ° s -8» O IO 7J . £ N c O r •«- re i . S £ - a i -re*-© - g © &| CO « * TJ •2 © (0 © N 1 2._ O) (0 © © re E ' JS ^ re ^ a ro") - -© . - ™ > ST 43 S CO o eg « C ri & * ™ S =» (J _ o> o u re^STJ CD — C . 5 43 ,x re « « w e SJ ^ co « "T © ° T- 2 .to 3 2 r.S 5 5, re a OiS £r o a. g ^ CN u_ >. co CO 3 co •Q E TJ 5 -8 co 1 to E 4? CO —I <0 CO c o w 8 . O l § s a o •c f 2 3" E w a. CL a to CL CL (0 .>< "to CO CO I E .5 o o c 3 O) O E © a. CO £ 3 CO w _> "•3 u 3 TJ O L. Q. CD a> n E C CO 3 co co ** «« o ° 3 t. as © u> CO 0) co E 5 re ^ 3^ CO > 0) TJ co co TJ CD — co re |2 » to .» o LU L -co O CD *" CO CO TJ CD l i ~ o > CO >:§ co si c c CD — X) *-i ° CD c TJ C -i > © £ • ^ ™ o ro to co oo •ej ? LU 3 E O 3 co 43 2 • . -E 'o -J o> CD -CO S - O S = 1 (I) CO O CO co *; m (0 C TJ .5? a. C CO • 3 O o « g.2 o * _ CO — ? o» >» «S co TJ fl) *» 3 i> 1 CD . (0 3 U o 00 £ g*(0 > CD , CD |= 00 CO o !_</>•«-O <D -* N Q. •£ CO Q. 0 CD CO E O..S w E <o a, « w := o c UN 3 •13 ^ O O Ol L Q.^ 10 £ co © — CO CO = o-o co £ n C CO CD s.s I 1 CO <J= ID <= I 3 _ •5>£? LU CO CO O w m o> JQ .E Ii c o co c co t: TJ O © i_ zz © 3 X! CO c CO C © 3 E = •!2 d +* a. 8<5§ El 5: "51 I ' I I £ °> Tom 3 >-§•2 .»£ — co •C o re n_ rf 3 O <= 2" i- 3 re © re o f = o = CO 00 eg Table 5 - Output from Canonical Correlation Analysis of data from O. nigrescens. Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: previous number of flower buds, previous plant diameter, snow-free date, thawing degree days, incoming global solar radiation and rainfall. Plant response variables are: prefloration time, plant diameter and number of flower buds. Prefloration is number of days from snow-free date to first flower open. Total observations were 23. Canonical Variate 1 2 Canonical Correlations 0.851 0.651 Eigenvalue 2.63 0.736 Canonical Cross-loadings Prefloration 0.339 -0.307 Plant diameter 0.735 0.321 Number flower buds 0.648 -0.222 Square root previous number of flower buds -0.032 0.420 Previous plant diameter 0.056 -0.025 Snow-free date 0.550 0.389 TDD -0.772 0.078 Log™ (rain) 0.619 -0.268 Radiation -0.557 -0.182 Axis 2 Figure 6 - Ordination biplot for O. nigrescens. Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Prefloration is number of days from snow-free date to first flower open. Total observations were 23. 29 For L decumbens, the C A N C O R analysis again resulted in two pairs of canonical variates (Table 6). The first pair indicated prefloration and floration were positively related to snow-free date and negatively related to rainfall, TDD and incoming global solar radiation (Figure 7). The second pair of canonical variates was redundant and offered no new information. Number of flowers was not well predicted by either pair of canonical variates. The variance explained by the first environmental canonical variate was 41%, that explained by the second was 17%, for a total of 58% of variability in plant response accounted for by the two environmental canonical variates. The C A N C O R analysis of data from V. Vitis-idaea resulted in two strong canonical correlations (Table 7). The first pair of canonical variates suggested that prefloration and floration were positively related to snow-free date and negatively related to incoming global solar radiation and TDD (Figure 8). Number of fruit showed the opposite relationships (i.e. positive relationship to radiation and TDD, negative relationship to snow-free date). The second pair of variates indicated that number of flowers and fruit were positively related to rainfall and previous number of flowers while negatively related to TDD. The first environmental canonical variate accounted for 70% of variability in plant response while the second accounted for 4%, for a total of 74% variability accounted for by the two environmental canonical variates. 30 Table 6 - Output from Canonical Correlation Analysis of data from L. decumbens. Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: snow-free date, rainfall, TDD and incoming global solar radiation. Plant response variables are: prefloration time, floration time and number of flowers. Prefloration is number of days from snow-free date to first flower open. Floration is number of days from first flower open to last flower dropped. Total observations were 57. Canonical Variate 1 2 Canonical Correlations 0.972 0.884 Eigenvalue 16.867 3.587 Canonical Cross-loadings Prefloration 0.513 0.751 Floration 0.966 0.087 Number flowers 0.120 0.061 Snow-free date 0.884 0.368 Rain -0.847 0.434 TDD -0.824 -0.469 Radiation -0.796 -0.507 Figure 7 - Ordination biplot for L. decumbens. Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Prefloration is number of days from snow-free date to first flower open. Floration is number of days from first flower open to last flower dropped. Total observations were 57. 31 Table 7 - Output from Canonical Correlation Analysis of data from V.Vitis-idaea. Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: previous number of flowers, square root rain and incoming global solar radiation. Plant response variables are: prefloration time, floration time, number of flowers and number of fruit. Prefloration is number of days from snow-free date to first flower open. Floration is number of days from first flower open to last flower dropped. Total observations were 23. Canonical Variate 1 2 Canonical Correlations 0.935 0.806 Eigenvalue 6.933 1.860 Canonical Cross-loadings Prefloration 0.889 0.057 Floration 0.819 0.317 Number flowers -0.143 0.414 Number of fruit -0.564 0.418 Previous number of flowers -0.049 0.490 Snow-free date 0.880 0.221 TDD -0.654 -0.577 Square root rain 0.236 0.767 Radiation -0.870 -0.256 Axis 1 # fruit ii „ # flowers radiation TDD 'sqrt rain snowfree date floration prefloration Axis 2 Figure 8 - Ordination biplot for V.Vitis-idaea. Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Prefloration is number of days from snow-free date to first flower. Floration is number of days from first flower open to last flower dropped. Total observations were 23. 32 The C A N C O R analysis of B. glandulosa data (excluding male reproductive phenology and number of catkins) resulted in two strong canonical correlations (Table 8). The first pair of canonical variates indicated that leaf-out was positively related to snow-free date, previous number of catkins and negatively related to rainfall. Number of female catkins appeared to have opposite relationships: negative relationship to snow-free date and previous number of catkins and a positive relationship to rainfall (Figure 9). The second pair of canonical variates showed that prefloration was strongly related to TDD and negatively related to snow-free date. The first environmental canonical variate accounted for 38% of variability in plant response, the second for 6%, for a total of 44%. Analysis of the B. glandulosa data (excluding the female reproductive phenology and numbers of catkins) resulted in two strong canonical correlations (Table 9). Snow-free date had a strong positive relationship while TDD was negatively related to leaf-out as was shown by first canonical variate (Figure 10). The first variate also showed that number of male catkins was negatively related to snow-free date and positively related to TDD. The second variate showed a weak negative relationship between male prefloration time and TDD (Table 9). A total of 48% variability in plant response was accounted for, 40% by the first canonical variate and 8% by the second. 33 Table 8 - Output from Canonical Correlation Analysis of data from B. glandulosa (female phenology and catkins). Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: square root previous number of catkins, snow-free date, log TDD, log rain and incoming global solar radiation. Plant response variables are: leaf out, female prefloration time, number of female catkins. Leaf-out is number of days from snow-free date to leaf emergence; prefloration is number of days from snow-free date to stigma receptivity. Total observations were 46. Canonical Variate 1 2 Canonical Correlations 0.973 0.919 Eigenvalue 17.477 5.466 Canonical Cross loadings Leaf -out 0.961 0.121 Female prefloration -0.146 0.908 Number of female catkins -0.607 -0.148 Square root previous number of catkins 0.433 -0.229 Square root snow-free date 0.606 -0.479 Logio (TDD) -0.435 0.006 Log10 (rain) 6.128 0.694 Radiation 0.257 0.332 female prefloration + log rain •4 radiation log TDD sqrt prev/oos_|_leaf-out i # female catkins + 4 ~^**==5|^ * catkins' ' snowfree~~~*- A x l s 1 date Axis 2 Figure 9 - Ordination biplot for B. glandulosa (female phenology and catkins). Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Leaf-out is number of days from snow-free date to leaf emergence; prefloration is number of days from snow-free date to stigma receptivity. Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Total observations were 46. 34 Table 9 - Output from Canonical Correlation Analysis of data from B. glandulosa (male phenology and catkins). Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: log previous number of catkins, square root snow-free date, TDD, rain and incoming global solar radiation. Plant response variables are: leaf-out, male prefloration, number of male catkins. Leaf-out is number of days from snow-free date to leaf emergence; male prefloration is number of days from snow-free date to pollen release. Total observations were 43. Canonical Variate 1 2 Canonical Correlations 0.981 0.852 Eigenvalue 24.926 2.654 Canonical Cross-loadings Leaf-out 0.975 0.054 Male prefloration -0.148 0.823 Number of male catkins -0.627 0.240 Log10 (previous number of catkins) 0.238 -0.139 Square root snow-free date 0.706 0.141 TDD -0.562 -0.347 Rain 0.048 0.023 Radiation 0.258 -0.106 # male catkins Axis 1 male prefloration snowfree date leaf-out Axis 2 Figure 10 - Ordination biplot for B. glandulosa (male phenology and catkins). Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Leaf-out is number of days from snow-free date to leaf emergence; prefloration is number of days from snow-free date to pollen release. Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Total observations were 43. 35 Three strong canonical correlations resulted from the C A N C O R analysis of data from female Salix spp. plants (Table 10). The first pair of canonical variates showed that leaf-out and female prefloration were positively related to snow-free date and negatively to TDD and radiation (Table 10). Number of catkins was related in the opposite way to these variables; negatively to snow-free date and positively to TDD and radiation (Figures 11 and 12). The second pair of variates indicated that leaf size and catkin length were related positively to rain. The third set of canonical variates showed that number of female catkins, leaf size and catkin length are moderately related to previous leaf size and previous number of catkins (Table 10). Total variance accounted for by the environmental canonical variates was 61%: 33% accounted for by the first, 16% by the second and 12% by the third. The C A N C O R analysis for male Salix spp. plants, resulted in two rather than three strong canonical correlations (Table 11). The first canonical pair showed that leaf-out and male prefloration were positively related to snow-free date and previous leaf size and negatively related to TDD and radiation (Figure 13). The second pair of canonical variates suggested that male prefloration was also related negatively to rain and previous number of catkins. The number of male catkins was not well predicted. The first two environmental and previous investment variates accounted for 52% and 14%, respectively, of variance in plant response for a total of 68%. 36 Table 10 - Output from Canonical Correlation Analysis of data from Salix spp (female phenology and catkins). Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: log previous number of catkins, snow-free date, TDD, rain and incoming global solar radiation. Plant response variables are: leaf-out, female prefloration, number of female catkins, leaf size and mature catkin length. Leaf-out is number of days from snow-free date to leaf emergence; prefloration is number of days from snow-free date to stigma receptivity. Total observations were 19. Canonical Variate 1 2 3 Canonical Correlations 0.981 0.889 0.778 Eigenvalue 26.258 3.752 1.538 Canonical Cross loadings Leaf-out 0.930 0.246 0.123 Female prefloration 0.857 -0.293 0.271 Number of female catkins -0.469 0.254 0.365 Leaf size -0.607 0.504 0.414 Catkin length -0.249 0.383 0.332 Square root previous number catkins -0.019 0.339 0.337 Previous leaf size 0.246 0.074 0.688 Snow-free date 0.924 -0.224 -0.002 Square root rain 0.114 0.803 0.129 TDD -0.960 -0.056 -0.150 Radiation -0.933 0.085 0.130 leaf size catkin length # catkins+ radiation\ TDD r sqrt rain leaf-out + snowfree date female A x l s 2 prefloratbn Figure 11 - Ordination biplot displaying axes 1 and 2 for female data of Salix spp.. Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Leaf-out is number of days from snow-free date to leaf emergence; prefloration is number of days from snow-free date to stigma receptivity. Total observations were 19. 37 Axis 2 previous le,)f length _ ; > # female catkin leaf " catkins length size sqrtfirevtous # catkins sqrt rain Axis 3 Figure 12 - Ordination bipiot displaying axes 2 and 3 for female data of Salix spp.. Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Total observations were 19. The analysis of data from S. tricuspidata showed two strong canonical correlations (Table 12). Plant diameter and number of flower stalks were related in a positive manner to previous plant diameter, previous number of flowering stalks and snow-free date and in a negative manner to TDD as shown by the first pair of canonical variates (Figure 14). This variate set also showed that prefloration is negatively related to previous plant diameter, previous number of flowering stalks and snow-free date and related positively to TDD. The second pair of canonical variates was redundant. The two environmental and previous investment canonical variates accounted for a total of 84% of variability in plant response of S. tricuspidata, the first accounting for 74% and the second 10%. 38 Table 11 - Output from the Canonical Correlation Analysis of data for male Salix spp. Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: log previous number of catkins, log mean temperature, TDD, rain and incoming global solar radiation. Plant response variables are: leaf-out, female prefloration time, number of female catkins. Leaf-out is number of days from snow-free date to leaf emergence; prefloration is number of days from snow-free date to pollen release. Total observations were 29. Canonical Variate 1 2 Canonical Correlations 0.980 0.934 Eigenvalue 24.367 6.875 Canonical Cross-loadings Leaf -out 0.932 -0.276 Male prefloration 0.860 0.447 Number of male catkins 0.254 -0.121 Leaf size -0.257 -0.373 Square root previous number of catkins 0.117 -0.318 Previous leaf size 0.581 -0.295 Snow-free date 0.896 0.292 Square root rain 0.207 -0.871 TDD -0.937 0.174 Radiation -0.910 -0.339 male prefloration Axis 2 Figure13 - Ordination biplot for male data of Salix spp. Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Leaf-out is number of days from snow-free date to leaf emergence; prefloration is number of days from snow-free date to pollen release. Total observations were 29. 39 Table 12 - Output from the Canonical Correlation Analysis of data for S. tricuspidata. Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: square root previous number of flower stalks, previous plant diameter, snow-free date, TDD, log rain and radiation. Plant response variables are: prefloration, number of flower stalks, plant diameter. Prefloration is number of days from snow-free date to first flower open. Total observations were 62. Canonical Variate 1 2 Canonical Correlations 0.947 0.846 Eigenvalue 8.768 2.509 Canonical Cross-loadings Prefloration -0.579 0.669 Number flower stalks 0.520 0.187 Plant diameter 0.876 0.316 Square root previous number of stalks 0.716 -0.075 Previous plant diameter 0.872 0.315 Snow-free date 0.306 -0.337 TDD -0.327 0.400 Log rain -0.052 0.021 Radiation -0.236 0.223 prefloration j q q Axis 1 ^ ^ ^ ^ ^ plant # flower stalks diameter ______— "previous plam i i Axis 2 _ J - — ^ . diamete'r previous # flower snowfree S(a/^s date Figure 14 - Ordination biplot for S. tricuspidata. Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Prefloration is number of days from snow-free date to first flower open. Total observations were 62. 40 The C A N C O R analysis of the male E. vaginatum data resulted in two strong canonical correlations (Table 13). Snow-free date, TDD, previous leaf length and previous number of inflorescence stalks were all positively related to leaf length and negatively to prefloration while rain influenced leaf length negatively and prefloration positively in the first canonical variate (Figure 15). The second variate showed that number of inflorescence stalks was positively related to rainfall, while negatively related to TDD, radiation and previous number of inflorescence stalks. A total of 68% variability was accounted for by the environmental and previous investment canonical variates: 54% by the first canonical variate and 14% by the second. The C A N C O R analysis of data for female reproductive structures from C. aquatilis resulted in two strong canonical correlations (Table 14). The first pair of canonical variates showed that leaf-out and to a lesser degree leaf length were positively related to radiation and TDD and negatively related to snow-free date and previous leaf length (Figure 16). Female prefloration showed the opposite relationships; negatively related to radiation and TDD and positively related to snow-free date and previous leaf length. The second pair of canonical variates showed that leaf length is also negatively related to rainfall. The first environmental canonical variate accounted for 20% of variability in plant response while the second accounted for 35% for a total of 55%. 41 Table 13 - Output from the Canonical Correlation Analysis of data from E. vaginatum. Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: square root previous number of stalks, square root previous leaf size, TDD, log rain and radiation. Plant response variables are: prefloration, number of inflorescence stalks, leaf length. Prefloration is number of days from snow-free date to time of pollen shed. Total observations were 39. Canonical Variate 1 2 Canonical Correlations 0.882 0.765 Eigenvalue 3.510 1.412 Canonical Cross-loadings Prefloration -0.800 0.241 Number inflorescence stalks -0.173 0.628 Leaf length 0.735 0.372 Square root previous number of stalks 0.564 -0.316 Square root previous leaf length 0.445 0.033 Snow-free date 0.823 -0.135 TDD 0.574 -0.572 Log™ (rain) -0.320 0.663 Radiation --0.065 -0.549 Figure 15 - Ordination biplot E. vaginatum. Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Prefloration is number of days from snow-free date to time of pollen shed. Total observations were 39. 42 Table 14 - Output from the Canonical Correlation Analysis of data from C. aquatilis (female phenology and inflorescences). Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: previous leaf length, snow-free date, log TDD, rainfall and incoming global solar radiation. Plant response variables are: female prefloration, leaf-out and leaf length. Prefloration is number of days from snow-free date to time of stigma receptivity. Total observations were 76. Canonical Variate 1 2 Canonical Correlations 0.923 0.917 Eigenvalue 5.778 1.073 Canonical Cross-loadings Female prefloration 0.709 -0.108 Leaf-out -0.844 -0.251 Leaf length -0.393 0.621 Previous leaf length 0.502 0.177 Snow-free date 0.729 0.292 Log,0 (TDD) -0.741 0.114 Rain -0.127 -0.612 Radiation -0.838 -0.103 leaf length Axis 1 + log TDD previous leaf length snowfree date ' 4-leaf-out radiation ^ •ain female prefloration Axis 2 Figure 16 - Ordination biplot for C. aquatilis (female phenology and inflorescences). Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Prefloration is number of days from snow-free date to time of stigma receptivity. Total observations were 76. 43 Two strong canonical correlations resulted from the C A N C O R analysis of data for male reproductive structures of C. aquatilis (Table 15). The first pair of canonical variates indicated that leaf-out and leaf length are positively related to radiation and TDD and negatively related to snow-free date and previous leaf length (Figure 17). This variate pair showed that male prefloration has the opposite relationships: negatively related to radiation and TDD and positively related to snow-free date and previous leaf length. The second pair of canonical variates implied that leaf length is negatively related to rainfall while male prefloration and leaf-out were positively related to rainfall. Total variability in plant response accounted for by the environmental canonical variates was 53% of variability; 24% by the first and 29% by the second. The direction of the multivariate correlations are, for the most part, consistent with the simple correlations (Table 16). Notable differences are the correlations between snow-free date and plant responses in B. glandulosa. The direction of correlation between growth measures and previous investment differed in the simple correlations for O. nigrescens and E. vaginatum. L. decumbens, Salix spp. and C. aquatilis showed no differences in direction between multivariate and simple correlations. In summary, snow-free date, TDD and incoming global solar radiation showed strong correlations with time to leaf-out in the three species for which there were data (Table 17). TDD proved to be the most common climatic variable correlated with prefloration time in the set of study plants. It had a negative relationship to flowering in O. nigrescens, pollen release in B. glandulosa and Salix spp., and 44 Table 15 -Output from the Canonical Correlation Analysis of data from C. aquatilis (male phenology and inflorescences). Data are from six years of annual sampling (1997 - 2002). Environment and previous investment variables are: previous leaf length, snow-free date, log TDD, rain and incoming global solar radiation. Plant response variables are: male prefloration, leaf-out and leaf length. Prefloration is number of days from snow-free date to time of pollen shed. Total observations were 75. Canonical Variate 1 2 Canonical Correlations 0.907 0.780 Eigenvalue 4.661 1.561 Canonical Cross-loadings Male prefloration -0.511 0.401 Leaf-out 0.812 0.349 Leaf length 0.462 -0.551 Previous leaf length -0.439 -0.258 Snow-free date -0.673 -0.354 Log10 (TDD) 0.728 0.011 Rain 0.065 0.565 Radiation 0.922 0.138 male prefloration + Axis 1 •4 leaf-out / rain radiation snowfree leaf length _ date i 1 log TDD + Axis 2 leaf length Figure 17 - Ordination biplot for C. aquatilis (male phenology and inflorescences). Points represent overall plant response. Arrows represent environmental and previous investment variables (labelled in italics). Points and tips of arrows are canonical loadings scaled by the eigenvalue of the particular canonical variate (axis). Only those variables with a cross-loading of greater than 0.3 (absolute value) on one or both axes are shown. Prefloration is number of days from snow-free date to time of pollen shed. Total observations were 75. 45 Table 16 - Simple correlations between plant response variables and environment and previous investment variables for the eight study species (a - h). Asteriks indicate where direction differs from multivariate correlations. a) 0. nigrescens b) V. Vitis-idaea Previous flower buds Previous diameter Snow-free date TDD Rain Radiation Prefloration -0.5012 -0.2635 0.1870 -0.3671 0.3287 -0.2255 Plant diameter 0.2176 0.0749 0.6467 -0.6254 0.4078 -0.5587 Flower buds 0.0352 * 0.2604 0.1800 -0.5983 0.5937 -0.2996 c) L. decumbens d) 8. glandulosa e) Salix spp. f) S. tricuspidata g) E. vaginatum ft) C. aquatilis Previous # flowers Snow-free date TDD Rain Radiation Prefloration 0.0307 0.8626 -0.6625 0.2689 -0.8546 Floration 0.0601 0.8355 -0.7982 0.5290 -0.8422 # Flowers 0.0227 -0.0803 -0.1926 0.4109 0.0561 # Fruit 0.4958 -0.3547 0.0432 * 0.2825 0.3303 Snow-free date Rain TDD Radiation Prefloration 0.7786 -0.0787 -0.8329 -0.8508 Floration 0.9146 -0.7994 -0.8652 -0.8412 # Flowers 0.1341 -0.0746 -0.1337 -0.1329 Previous # catkins Snow-free date TDD Rain Radiation Leaf out 0.3796 -0.5381 * -0.4216 0.1913* 0.3211 Female prefloration -0.2777 0.5653 * 0.0662 0.6850 0.2721 # Female catkins -0.4164 0.2759 * 0.3423 -0.4572 0.0211 Male prefloration -0.2238 -0.0197* -0.2965 0.0802 -0.1888 # Male catkins -0.0116 0.3766 * 0.4167 -0.2439 -0.0343 Previous # catkins Previous leaf size Snow-free date Rain TDD Radiation Leafout 0.1372 0.3634 0.8144 0.3488 -0.9490 -0.8406 Female prefloration -0.0045 0.4359 0.8841 -0.1267 -0.8726 -0.8016 Male prefloration -0.0517 0.3567 0.9207 -0.2294 -0.7340 -0.9590 # female catkins 0.0948 0.2288 -0.5047 0.2268 0.3824 0.5177 # Male catkins 0.1075 0.2251 0.1306 0.2338 -0.2166 -0.1630 Leaf size 0.4182 0.2633 -0.6963 0.4473 0.4803 0.6922 Mature female catkin length 0.3351 0.3157 -0.3017 0.3206 0.1514 0.2960 Previous # stalks Previous diameter Snow-free date TDD Rain Radiation Prefloration -0.4950 -0.2830 -0.4590 0.5204 0.0503 0.3263 # Flower stalks 0.4493 0.5680 -0.1136 * 0.0763 * 0.0417 0.1418 Leaf length 0.6403 0.9262 0.1380 -0.1378 -0.0345 -0.1149 Previous # stalks Previous leaf length Snow-free date TDD Rain Radiation Prefloration -0.6187 0.4163 * -0.8339 -0.7068 0.4605 -0.2090 # Inflorescence stalks -0.3568 0.0148 -0.1862 -0.5712 0.6809 -0.2569 Leaf length 0.3094 -0.3640 * 0.5768 0.1948 0.0186 -0.4102 Previous leaf length Snow-free date TDD Rain Radiation Male prefloration 0.1412 0.2532 -0.3262 0.0831 -0.4596 Female prefloration 0.3897 0.5396 -0.5320 -0.0980 -0.6777 Leaf out -0.5087 -0.7611 0.6553 0.3127 0.7895 Leaf length -0.0203 -0.0487 0.4223 -0.4986 0.2596 46 w O.S2 TJ *S ' £ S .E i . E .2 o E i s | c *• "= o © to •> 0 •o •a Urn <0 TJ 10 w O |> E 2 5 ° 2 8 co «J _ a> to O 3 CO TJ *- 3 C . a * , o ~ to r-w Q. O I o *8 o — o 2 •8 E co to CD « .2 CD 2 CO _ £ . § . > • o) > *• co * o -2 ^ 5 CD U- t "5 to o co ~ o CO *• CO — ^ 3 .2 c 3 O > O Ox!, n c - • co c co 5)v co 0 ^ to o> " CM Jr 3 0)0 O p C ° © £ .E co £ ** r i CO _ 3 ? g C Igll _ CO — jS m ® ° © 8 ^S-o #£ »— » -1 CO <1) Q) r» O o > co 2 CD C CD fl) CO S O t. H E c u 2 ta. o TJ 1 2 CD .E >-| * 6 2 LL 0) CO c © o .E £ £ E g © 2 © ° E c 2 i 0 © •E > c o Q.-XJ CO co c o E = o o *3 CO k-o «C © 2 E OL V © CO o X I CD CO > 5 LU "O CO T J CD CU X I s i (U CO > £ 111 T J co > cu CO CO X J cu CD 5 o c: CD +i CO CO to 0 CO © co E .© © CO E c © o .E S £ <-> c co a co E s= > H O T ) Q Q © ^ C CO ** •E Q © 3 h 2 J O CO © © a to CO X J CD CD CO CD 00 CD +1 ai co X J CD CD o TO +1 CO CD X I CO 3 H "9 t O CO CD 5 Q T J CO e CD > X I CO ZJ O CO CD 5 Q X J CO E c CD OJ o>| © E <u CD X J CD CO CO c 1 CD > TO CO Q Q o <°. CN CO +1 CM CM ™ CM TO 10 c CD O <0 c TO CO I TO "£ 10 c CD •Q i o CO X J CD CD I— I o a CO CO -H CO I c JO O) CQ CD CN ID -H CO LOP CM 00 +1 CD O CD LO 8-to CD CO CD CM TO 3 I to I 2 LU CD CD X J CU CO a a X J CO +1 CD CD CO LO c o X J TO +1 .eg S I CO O stigma receptivity in C. aquatilis. TDD was positively related to prefloration time in S. tricuspidata. Snow-free date was the next common climatic variable correlated with prefloration time. L. decumbens and V.Vitis-idaea, both aestival flowering plants, showed positive relationships to snow-free date while prefloration in E. vaginatum was negatively correlated with snow-free date. Incoming global solar radiation appeared to be negatively related to pollen release in C. aquatilis and stigma receptivity in Salix spp. Stigma receptivity in 8. glandulosa was positively correlated with rain. Lastly, snow-free date showed highest correlation with floration time in the two species for which there were data. For the five species that had growth measurements, two were strongly related to TDD (one positive and one negative), one species was most highly correlated with previous growth, another with snow-free date and the last with incoming global solar radiation (Table 18). Reproductive investment in two species was strongly related to rainfall, another two to TDD, snow-free date and previous growth measure were the variables most strongly related to reproductive investment in the last two species. There did not appear to be any patterns according to flowering strategy or growth form. In most cases, previous growth and reproductive investment were not strongly related to current plant responses in the presence of the climatic variables; however, there appeared to be some interesting underlying relationships (Table 19). 8. glandulosa, and C. aquatilis showed evidence of cyclic resource allocation 48 E o o .E > >- CO 5 2 1 o ro c i > s i °_ OT E?t CD CO > 5 III "D s i OT m CO > 5= U J x> x> OT 3 O CO a> <: Q -o CO c I > to 3 S £ 2 1 O CO <u «» Q -o CO c 1— CD > CO E c CD M CO > OT CD CO CO cn T3 CD CD > CD cn "O CD OT c 2 c o a.*: o .£ S o E 0 (0 E Q Q I-<D jo a> o o a3 5 CD , ™ O CD C 0) OT •<= a CD i s C CO OT "O X J __ CD a> co 5 CD i s JO a. O CD "II CL "O CO CD 3 ~ o c ^ jo (0 ce <u CD XI Z XI OT O o co co XI cu CD o | £ x : 1 O o i- c 0) CO 1* Z CL OT c CO o o o i c i co o XI O o i -z 2. wer *—• o o wer OT M — pla nee tus o 1 CD per O i CD scei per XI OT X> CD OT E E O _£ CO 3 M — CO Z •*-» OT Z c OT c 0) — c to £ g> to l l II i i CD 2 m E co C M C M CO co • +• C O C O 0) CO E C D i +• CO + CO CM | -c CD u » i E > 2 Q. O -O Q Q 3 (0 (0 CD S CD ^—' JO CO 0- 73 I ? + CD od C M I O C M 0) o '5 Q> Q. CO g CD O 10 ..I O 5 S .A C M in + CO + C O C O + OT Q Q 3 O > c: E £ ro co Q- O.TJ CD CD OT T3 O ro ro CD N ro CD CD •4-» 5 E jo ro CL T3 + CO o ^ co r-co ^~ ~- in co - i C M + O , C D ; C O i O C D | C M • (0 41 c . JS CQ O) Q. a. CO to to I CO £ I (0 •I ui2 .<2 I co o CO t o CO 0 O 0 Q. CO ro Q. O s e g '» O) E » > co 3 0 CO > 4-+3 O Mt TJ CO O 0) C >, a. 0 •— »- CO *§ « 2 1 2 o « « a 2 • >Z CO CO < § = o.2 15 2 2 c o 0 u 0 — 28 0 •= •° o 2 = Q. n _E o » J= I i ' 13 2 0 £ 1 0 cn £ « E O ka — 0 CO 0 r— TJ i f o 2 H to re 2 k_ o o CO re 0 E 0 > u 3 TJ O i -Q. 0 DC CO 0 o 0 Q, (0 xi i o ro cz i— CD > CO i CD CO d o CO •o o cz > CD ro (0 c 0 I c ro •o c: CU 0) u-CO J_ 5 1 LU to CO _> ~-t—' CO CD CO CU o o CZ CO 3 g '> CD t • ro "D CO 3 O 3 T3 XI 6 B CD JE Q «5 ro c I— CD > CO c .—, CD ca ro E d CD M — t ro 3= T3 CO 3 O 3 "O XI CD j= Q » ro c CD > DJ e CD CD — N C S _£ •{S CO ro oj O - J c _£ -*—' ro o d cz CO ,—. 3 CD .2 ro £ E *-a. ^ CO CD CZ c c; CD CD t= — N CO c -t—1 m O TO m o • ro £Z CO 0 s ro c: CO CO O c co O) 00 CO c 4—' ro o d tz CO 3 g '> cu a. " I S i 3£ c ro E c a> CD O) a>.E a> E .?..P ro > CO CD ro CO ro CD +-» CO cu % -o -a "= c d J5 CZ o ro t o cu 1 cz g 2 3 o ? cu ro j= CD 0 I cu -»—• cu 1 1 o "° co „ __ c d ^ "co 0-CD cu N N CO CO»*— M — ro ro a> a_ CO CO 3 3 O O ' > '> a> CD DL CL Q. 5 § CO CO cu g: o o c CO 3 g "> cu cu cn "O cu CO co c i CD > cn CZ CO cu cu o CZ CU o CO cu 1 o **— CZ d CZ CO 3 o CO •> cu ro CL Vi CZ o ro i o CU I CL c ro CL O CD > E £ ro CL T3 I _z CO ro a> in 3 g > cu s a CO c & § ui cu CD TJ CU CO ro > CO CD CD ro o in x: •t—i 3 O leng ro ro CP cu _i _i cn cz a_ H — ro a> CO 3 g '> cu <o CD a CO patterns. Previous investment in reproduction was negatively correlated with current reproductive investment, for the former species, and previous investment in growth was negatively correlated with current investment in growth, for the latter. Previous investment also appeared to be related to phenology in some cases. Large investment in growth during the previous season was correlated with increased time to flower in Salix spp. (male) and C. aquatilis, and contracted time to flower in S. tricuspidata. The previous growth investment correlations for E. vaginatum were not interpreted as they were inconsistent with those of the simple correlations. Previous season's reproductive investment had a negative relationship to prefloration in four species: two vernal flowering species, O. nigrescens and Salix spp. (male) and two aestival species, E. vaginatum and S. tricuspidata. 2.4 Discussion Over the six years of annual sampling, the study plants showed a large range of variability in prefloration phase length with the vernal flowering species being more variable in time to flower than the aestival flowering species. This difference in variability with flowering strategy does not appear to be linked to variability in snow-free date as vernal flowering plants showed the strongest relationship to TDD while for aestival flowering species, strong relationships were evident for both TDD and snow-free date. This differs from what has been suggested in the literature (i.e. that the phenology of vernal species are more sensitive to timing of snowmelt than aestival species) (Molau 1997b). The interaction between snow-free date and prefloration is of interest. The literature again supports the pronounced influence of snow-free date on arctic plant 51 phenology (Kudo 1991, Galen and Stanton 1991, Inouye and McGuire 1991, Woodley and Svoboda 1994, Walker et al. 1995, Molau 1996, Inouye etal. 2002). However, many of these studies examined actual dates of events rather than the interval from snowmelt to flowering. In this study, variability in the actual date of flowering was on the same scale as variability in snow-free date implying strong correlations between the two. Yet, the date of flowering for some species was more variable and for other species less variable than snow-free date, suggesting the influence of factors other than snow-free date. It is less obvious how snow-free date is related to the length of the prefloration phase in arctic plants. Snowmelt is obviously an important trigger in the Arctic, signalling favourable conditions for plant growth (e.g. increased light and temperature). However, the relationship between snowmelt and spring phenology is not straightforward. In this study, the suite of eight low arctic plant species showed both positive and negative correlations between length of prefloration phenophase and date of snowmelt. However, positive and negative correlations do not easily translate into earlier and later than average flowering dates. For example, those plants with a positive relationship between prefloration and date of snowmelt will flower earlier with early snowmelt. However, those plants with a negative relationship could flower earlier or later than average depending on how much earlier snowmelt occurred and how much longer the prefloration phase became. The mechanisms through which snow-free date exerts its influence on arctic plants are not clear. A negative correlation between date of snowmelt and 52 prefloration phase length may be due to the removal of the physical protection from extreme temperature events that plants receive by being snow covered (Inouye et al. 2002). If snowmelt occurs early, then plants are subject to quite cool temperatures and floral development takes longer (Inouye and McGuire 1991). When snowmelt occurs later in the spring, temperature extremes might be less likely to occur and, therefore, development can occur quickly and time to flower is reduced. In his extensive examination of arctic plant phenology in Greenland, Sorenson (1941) corroborates this idea. He states that "...the prefloration time will as a rule be the shorter the later the melting of the snow takes place, since the rise of temperature will be the more rapid" (1941: 68). Two of the evergreen study species, L. decumbens and V. Vitis-idaea, showed positive correlations between prefloration and snowmelt; with earlier snowmelt, time to flower was contracted. In these cases, it is likely that earlier snowmelt allows increased absorption of radiation by evergreen leaves thus increasing rates of bud development. This has been observed for species with photosynthetic nutlets, such as Ranaunculus nivalis, where early snowmelt speeds up bud development through radiative warming, even when soils and roots are frozen (Molau 1997a). Another hypothesis is that timing of snowmelt influences timing of soil moisture and nutrient availability (Walker et al. 1995), thereby affecting plant development and growth. Among the study plants, TDD was the climatic variable most commonly related to prefloration time, consistent with the findings of Bean and Henry (2002) for low arctic plants and Thorhallsdottir (1998) for the majority of 75 plant species in 53 central Iceland. In most cases, the literature shows that phenology advances with increased temperature (Arft et al. 1999, Sandvik and Totland 2000) and the relationships evident from this study appear to support this. However, one of the study species, S. tricuspidata, showed a positive correlation between prefloration length and TDD such that flowering was delayed with greater accumulated heat. This species also showed later flowering in years of early snowmelt. It may be possible that lack of moisture was the issue here. At this location, S. tricuspidata inhabits a dry, raised beach ridge and perhaps late snowmelt and cool temperatures provide appropriate soil moisture levels for rapid development of buds; in years of early snowmelt and high temperatures these conditions are not created and the plant is late to flower. As soil moisture levels were not measured there is no direct evidence for this theory. In two species, C. aquatilis and Salix spp., flowering phenology was most strongly correlated with radiation. This is consistent with the literature and is, again, thought to act through soil warming (Molau 1997b, Marsden 1992) and/or through the warming of buds and other photosynthetically active structures (Molau 1997a). In the case of Salix, soil warming might be a plausible explanation as it is found in an open-canopy habitat. However, C. aquatilis is found in the wet lowlands and it doesn't seem that radiative soil warming could take place in this location. Again, because soil and bud temperatures were not measured these relationships could not be tested. Many of the study plants did not show strong relationships between climate and phenology. Four of the eight study species, O. nigrescens, L. decumbens, B. 54 glandulosa and S. tricuspidata, had correlations of less than 0.6 between prefloration phase length and the suite of environment and previous investment variables suggesting that variables other than those included in the analysis are influencing flowering. It has been suggested that plants at their latitudinal extreme will not be as responsive to changes in climate as their southern counterparts (Sparks et al. 2000) nor will stressed ecotypes (Fetcher and Shaver 1990). These may be possible explanations for the absence of strong correlations between climate and phenology in the study species. There were no obvious patterns according to growth form or flowering strategy in variability of growth responses among the study plants nor between growth and climatic variables. The climatic variable most correlated with annual growth appears to be quite species specific as has been reported previously (e.g. Chapin and Shaver 1985). In this study, O. nigrescens and Salix spp., showed strong correlations between growth and TDD, negatively and positively, respectively. A positive correlation between TDD and growth has been shown in the literature (Walker etal. 1995, Totland 1999, Pop etal. 2000), whereas, a negative response is not documented. Snow-free date was most correlated with growth for only E. vaginatum, in a positive manner (later snowmelt more growth). Walker et al. (1995) reported a similar result with Bistorta bistortoides in an alpine setting. They hypothesized that in late snowmelt years, optimum moisture and soil nutrient conditions coincided such that growth was enhanced. Variability among the study plants in annual reproductive investment was larger in vernal species as compared to aestival. This may be due to the fact that 55 reproductive structures of the vernal species are more susceptible to damage in the spring with variability in snow cover, frost and other extreme events (Molau 1993). After these influences have been factored in, the count of reproductive structures in the spring may not, therefore, accurately reflect investment made the previous season (Inouye and McGuire 1991). Thawing degree days was the most common variable correlated with reproductive investment in the vernal species but showed no pattern with the aestival species. Snow-free date showed both positive and negative relationships to reproductive effort which is consistent with the literature (Kudo 1991). The underlying patterns in growth and reproductive investment were perhaps the most interesting results to come out of this research. B. glandulosa showed alternations in annual allocation to catkins while C. aquatilis showed patterns of high investment in growth followed by low investment the subsequent year. These results are consistent with the findings of Johnstone and Henry (1997) for Cassiope tetragona, where plant responses to climate are mediated or constrained to greater or lesser degrees by internal patterns of resource allocation. Correlations between previous season's reproductive investment and spring phenology in four of the study species, O. nigrescens, Salix spp., E. vaginatum and S. tricuspidata, is noteworthy. Perhaps the production of more flowers results in less differentiation such that development time is increased the following spring. 2.4 Conclusions Plants respond in a complex manner to their environment and appear to be somewhat constrained by internal patterns in resource allocation. This study of eight 56 low arctic plant species showed that temperature and snowmelt are the climatic variables most correlated with spring phenology, whereas temperature, rainfall and snowmelt were most correlated with growth and reproduction. Some generalized patterns in plant responses could be made according to flowering strategy; however, growth form offered little utility in this regard, although the eight study species did not offer much duplication in growth forms. In light of these results, plant responses to climate change in the Arctic will be difficult to predict, as they will depend on the particular combination of climatic conditions to which the plants are exposed. Clearly, warmer temperatures decrease time to flower in most species, yet this response is complicated by snowmelt and precipitation patterns. Growth and reproductive responses were complex with no clear pattern among species. What becomes obvious from this research is that because of differential, species-specific responses to climate, the vegetation community of the low arctic will change with altered climatic patterns as has been confirmed in other studies (e.g. Chapin and Shaver 1985, Chapin etal. 1995). These results also speak to the necessity of comprehensive, long-term ecological research in the Arctic. This analysis of six years of data on eight low arctic plant species showed not only phenological, growth and reproductive relationships to climate but also revealed some underlying influences of plant allocation patterns. Although these results are useful, as more data on more species are collected, the analyses simply become more meaningful. Clearly, teasing out individual plant responses to a suite of climatic variables that are shifting 57 in complex ways requires large datasets covering a wide range of species over a long time period. To this end, long-term comprehensive datasets are a necessity. 58 C H A P T E R III - EXPERIMENTAL WARMING 3.1 Introduction Experimental warming of arctic plants leads almost exclusively to the advancement of spring phenology (Alatalo and Totland 1997, Jones et al. 1997, Levesque etal. 1997, Molau 1997a, Stenstrom and Jonsdottir 1997, Welker etal. 1997, Arft et al. 1999). Growth and reproductive responses, however, do not show consistent responses (Chapter 1). In many cases, growth increased with warming in the first years of experimentation and shifted toward increased reproductive effort in later years suggesting the possibility of primary and secondary responses (Arft ef al. 1999). There is, however, evidence for improved reproductive effort and success in the initial phase of the experiment (Totland 1999). Although arctic plant phenology seems quite responsive to warming, there is some discrepancy in the literature regarding growth and reproductive responses of these plants. It has been hypothesized that plants may not be exhibiting primary and secondary responses to warming but rather alterations of allocation as internal resources are utilized and exhausted (Johnstone and Henry 1997). How these patterns are altered with changing phenologies as climate slowly warms will be crucial to the success of individual species in the future. In this light, further examination of growth and reproductive responses of arctic plant species under scenarios of climate warming is essential. This chapter presents the experimental approach and results of the simulated warming of three low arctic plant species, Ledum decumbens, Vaccinium Vitis-idaea and Eriophorum vaginatum, at the Tundra Ecosystem Research Station, Daring 59 Lake, Northwest Territories. Phenology, growth and reproductive investment are examined over two years of manipulation. 3.2 Methods Open-topped chambers (OTCs) were used to simulate climate warming on three low arctic plant species, L. decumbens, V. Vitis-idaea and E. vaginatum in two habitats, lichen heath (for the former two species) and tussock tundra (for the latter). Hexagon-shaped OTCs, 1.5 m in diameter at the base, were constructed with Lexan greenhouse plexiglass in accordance to ITEX specifications (Marion 1996). Four OTCs were situated around each 10 X 10 m control plot and five 'individuals' were permanently marked for each species in each of four OTCs for a total of 20 plants. Twenty plants were marked in the control plot. Each OTC was considered the experimental unit because it is the smallest unit to which a treatment is randomized (Hurlbert 1984). In other words the replicates within the OTC are sub-samples and the mean values for each OTC were used in statistical analysis. As a result, there were only four replicates in the experimental treatment and therefore, the larger control plot was subdivided into four subplots for comparison with the treatment. Air temperature at 15 cm above ground surface and soil surface temperature at 2 cm below ground surface are recorded within the OTCs and control plots in each habitat type using automated data-loggers from date of snowmelt until August 15 (2001) and August 28 (2002). The phenological, growth and reproductive variables measured in the OTCs and control plots on L. decumbens, V. Vitis-idaea and E. vaginatum are given in Table 1 of the previous chapter. 60 The method of data collection was the same as that described in Section 2.2.3. 3.2.1 Data analysis Air and soil temperatures were analysed using a split-plot Analysis of Variance. Because of the nonlinear nature of temperature over time, quadratic terms, time-squared and time-cubed were included in the model. A partial F-test was used to compare the model with temperature as a function of time only to a model including treatment, year and interactions. Partial F was calculated as: [SSE(reduced) - SSE(full)]/rr / M S E (full) where S S E = sum of squares error M S E = mean square error rr= # restrictions from full down to partial. Residuals were tested for normality and homogeneity of variance. Where these conditions were not met data were rank transformed and normalized. The Type III Sum of Squares was interpreted as there is no clear order of effects. Plant response data were also analysed using a split-plot ANOVA. Each variable was analysed separately as missing values precluded the use of a multivariate technique (i.e, sample size was too drastically reduced when data were pooled). Actual dates of phenological events were analysed as were phenological phases. Because of limited sample sizes, two phenological events were selected for analysis along with a measure of growth and reproduction. Phenological phases were calculated such that prefloration and leaf-out are the number of days from snow-free date to the date of the phenological event (onset of flowering, and green-61 up, respectively). Floration is number of days from first flower open to first flower shed. Residuals were tested for normality and homogeneity of variance. Where these assumptions were not met data were rank transformed and normalized. The error term to test for treatment effects was adjusted to reflect the sub-sampling within chambers. Again, Type III sum of squares was interpreted. All analyses were done using Proc G L M and Proc Rank in the computer software package S A S , version 8. 3.3 Results The partial F-tests to determine whether air and soil temperatures differed between controls and OTCs showed significant differences in every case (a = 0.05 was used for all tests) indicating that the full model of treatment and year effects better represented the data than just time alone (i.e. treatment effects were evident) (Table 20). Figure 18 clearly shows that OTCs warm predominantly from 12 noon to 7 pm while, at other times of the day, there is no obvious warming effect; these trends differed slightly with year. Generally, the OTCs increased mean daily air temperature in both habitats by approximately 0.5 °C in both study years and increased mid-day maximum temperature by approximately 1.5 °C in the lichen heath and by approximately 0.75 °C in the tussock tundra. There was some variability among years. Daily mean soil temperatures were increased by approximately 0.5 to 1°C in both habitat types (Table 21). Difference in mid-day maximum soil temperature between OTCs and control was quite variable among years and habitats. Mid-day maximum soil 62 temperature in lichen heath habitat in 2003 was double that of 2002. In the tussock tundra mid-day maximum soil temperature was actually 0.5 °C cooler in the OTC than in control plots. Table 20 - Comparison of full and partial models for temperature x time relationships: a) air temperature in lichen heath habitat; b) air temperature in tussock tundra; c) soil in lichen heath habitat, and; d) soil temperature in tussock tundra. Partial F is calculated as [SSE(reduced) - SSE(full)]/rr / WISE (full) where rr= # restrictions from full down to partial. b) Model Source df Slim of Squares Mean square F partial F crlt(8,11,0.95) Full model error 11 29027 394361.5 758725.8 35851.0 26.1 26.9 3.35 Partial model error 3 29035 388727.6 764359.6 129575.9 26.3 Model Source df Sum of Squares Mean square F partial F crit(8,11.0.95) Full model error 11 29249 378839.7 740449.0 34440.0 25.3 13.8 3.35 Partial model error 3 29245 376054.7 743234.1 125351.6 25.4 c) d) Model Source df Sum of Squares Mean square F partial F crit (8,11,0.95) Full model error 11 29027 9347.3 19685.1 849.8 0.678 145.6 3.35 Partial model error 3 29035 8557.6 20474.8 2852.5 0.705 Model Source df Sum of Squares Mean square F partial F crlt (8,11,0.96) Full model error 11 29237 9131.0 20112.0 830.1 0.688 104.0 3.35 Partial model error 3 29245 8542.0 20701.1 2847.3 0.708 63 Table 21- Difference in daily mean and mid-day maximum air and soil temperature between OTCs and control plots in lichen heath and tussock tundra habitat. Data are means ± standard deviation of 68 days in 2002 and 83 days in 2003. Year Lichen heath Tussock tundra Air temperature Soil temperature Air temperature Soil temperature Difference in mean temperature 2002 0.6 ± 0.6 °C 0.5±1.4°C 0.4±0.5°C 0.5 ± 0.1.1 °C 2003 0.6 ± 0.9 °C 1.0±1.9°C 0.4 ± 0.4 °C 0.8±1.9°C Difference in Mid-day maximum 2002 1.7 ± 1.6 °C 1.4±2.3°C 1.0±1.7°C -0.5 + 2.4 °C 2003 1.4±1.5°C 3.6±4.7°C 0.4 ±0.7°C 1.9 +3.6 °C The full model for the air temperature / time relationship in lichen heath habitat showed there was no overall treatment effect (Table 22a). However, because the interaction terms between treatment and the time quadratic terms were significant we can conclude that treatment effect varies with time. Year had a significant effect on temperature and how it influenced the temperature/time relationship depended on time. The analysis of the air temperature/time relationship in tussock tundra was very similar to that of lichen heath (Table 22b). There was no overall treatment effect, but the effect of treatment varied with time. In this case, there was no overall year effect, but the effect of year on temperature among treatments varied with time. Soil temperatures in both habitats showed analogous results (Tables 22c and d). Temperature differed significantly between the control and OTCs and among years. Also, because all the interaction terms were significant, the treatment and year effects varied with time. Again, this is evident in Figure 18. 65 Table 22 - Analysis of Variance output from the comparison of air and soil temperatures between control plots and OTCs over two years: (a) air temperature in lichen heath habitat (n= 29 039); (b) air temperature in tussock tundra (n= 29 249); (c) soil temperature in lichen heath habitat (n= 29 039), and; (d) soil temperature in tussock tundra (n= 29 249). a) Source df F-value Pr> F Time 1 427.16 <0.0001 Timesq 1 2454.08 <0.0001 Timecu 1 3823.42 <0.0001 Treat 1 0.75 0.3860 Time*treat 1 0.26 0.6088 Timesq*treat 1 2.96 0.0853 Timecu*treat 1 9.26 0.0023 Year 1 4.07 0.0437 Time*year 1 2.71 0.0997 Timesq*year 1 12.34 0.0004 Timecu*year 1 19.41 <0.0001 c) Source df F-value Pr> F Time 1 1481.5 <0.0001 Timesq 1 3656.5 <0.0001 Timecu 1 4550.8 <0.0001 Treat 1 103.9 <0.0001 Time*treat 1 40.2 <0.0001 Timesq*treat 1 41.4 <0.0001 Timecu*treat 1 32.7 <0.0001 Year 1 29.6 <0.0001 Time*year 1 9.1 0.0026 Timesq*year 1 66.6 <0.0001 Timecu*year 1 113.8 <0.0001 b) Source df F-value Pr> F Time 1 408.65 <0.0001 Timesq 1 2359.21 <0.0001 Timecu 1 3620.42 <0.0001 Treat 1 0.09 0.7633 Time*treat 1 0.55 0.4599 Timesq*treat 1 5.02 0.0250 Timecu*treat 1 9.21 0.0024 Year 1 0.54 0.4606 Time*year 1 7.05 0.0079 Timesq*year 1 12.33 0.0003 Timecu*year 1 17.07 <0.0001 d) Source df F-value Pr>F Time 1 2665.5 <0.0001 Timesq 1 4577.8 <0.0001 Timecu 1 4867.8 <0.0001 Treat 1 41.3 <0.0001 Time*treat 1 8.2 0.0041 Timesq*treat 1 0.5 0.4703 Timecu*treat 1 8.4 0.0037 Year 1 35.3 <0.0001 Time*year 1 0.2 0.6661 Timesq*year 1 28.7 <0.0001 Snow-free date in the L. decumbens plots did not differ between treatments (Table 23). However, flowering phenology, in terms of the actual date of event, did show significant differences between control and OTC and the magnitude of these differences depended on year (Figure 19). The length of the phenological phases, prefloration and floration, were not significantly different among treatment, but were among years (Table 23). There was a significant difference in amount of branch growth between control and OTC in 2001 but not 2002 (Figure 20a). Number of fruits did not differ between treatments (Figure 20b). 66 Table 23 - Output from Analysis of Variance on plant responses in L. decumbens from control plots and within OTCs. Year effects and the interaction between treatment and year were analysed when sufficient data were present. The variable snow-free date was rank transformed prior to analysis. Prefloration is number of days from snow-free date to time of flower open. Floration is time from flower open to flower shed. Variable Sample size Source F-value P-value Snow-free date 44 Treatment 5.44 0.1020 Year 20.05 <0.0001 Flower open 69 Treatment 13.06 0.0364 Year 110.92 <0.0001 Treat*year 5.30 0.0249 Flower shed 69 Treatment 15.75 0.0286 Year 181.33 O.0001 Treat*year 7.29 0.0090 Prefloration 40 Treatment 1.67 0.2872 Year 2.92 0.0977 Floration 69 Treatment 0.06 0.8197 Year 5.02 0.0288 Treat*year 0.02 0.8861 Number of fruit 69 Treatment 0.46 0.5476 Year 0.62 0.4330 Treat*year 1.1 0.2995 Growth increment 80 Treatment 7.05 0.0767 Year 1.47 0.2287 Treat*year 4.73 0.0330 205 135 -I — — — — , , . snow-free date flowers open flowers shed Figure 19 - Flowering phenology of L. decumbens in OTCs and control plots. Data points are means of 4 replicates for all phenological events in OTCs for both years. They are means of 4, 18 and 18 replicates for control 2001 and means of 20,19 and 19 replicates for control 2002 for the phenological events of snow-free date, flowers open and flowers shed, respectively. Error bars are ± 1 standard deviation. No data exists for snow-free date for OTCs in 2001. Asteriks indicate significant difference in overall means as well as significant effect of year on treatment. 67 25 20 | 15 | 10 E 5 0 II a) 1 control OTC 2001 control OTC 2002 2001 2002 mean annual branch growth 25 20 15 10 5 0 n=18 n=4 n=20 n=4 I I I I control OTC 2001 control OTC 2002 2001 2002 mean annual number of fruit b) Figure 20 - Annual growth increment (a) and reproductive investment (b) in L. decumbens. Sample sizes for mean branch growth are 20 for controls, 4 for OTCs. Sample sizes are as indicated for mean number of fruit. Error bars are ± 1 standard deviation. Lower-case letters indicate significant differences. The analysis of the V. Vitis-idaea plant responses to simulated warming (Table 24) showed that none of snow-free date, date flowers open or date flowers shed differed from the control (Figure 21). The prefloration phenophase, however, was significantly shorter in OTCs. There was also a significant difference in reproductive investment (number of flowers) between control and O T C in 2001, with fewer flowers in the OTCs (Figure 22a). The analysis showed that although there was no overall significance in growth between control and OTC, year significantly affected the treatment effect (Table 24). Figure 22b shows that in 2001, growth in the chambers was higher than that of control plots whereas in 2002 it was lower. 68 Table 24 - Output from Analysis of Variance on plant responses in V.Vitis-idaea from control plots and within OTCs. Year effects and the interaction between treatment and year were analysed when sufficient data were present. The variables snow-free date, flower number and fruit number were rank transformed prior to analysis. Prefloration is number of days from snow-free date to time of flower open. Floration is time from flower open to flower shed. Variable Sample size Source F-value P-value Snow 44 Treatment 7.63 0.0700 Year 20.49 <0.0001 Flower open 45 Treatment 0.98 0.3961 Year 87.91 <0.0001 Treat*year 1.10 0.3021 Flower shed 43 Treatment 0.80 0.4366 Year 43.65 <0.0001 Treat*year 0.61 0.4397 Prefloration 33 Treatment 13.26 0.0357 Year 0.11 0.7474 Floration 43 Treatment 0.01 0.9315 Year 0.04 0.8410 Treat*year 0.04 0.8410 Number of flowers 55 Treatment 39.55 0.0081 Year 78.69 <0.0001 Number of fruit 71 Treatment 1.94 0.2583 Year 21.16 <0.0001 Treat*year 1.59 0.2121 Growth increment 80 Treatment 1.54 0.3032 Year 3.63 0.0608 Treat*year 10.63 0.0017 230 220 210 200 1 E 190 3 >> 180 ro •o 170 160 150 140 •control 2001 •control 2002 OTC 2001 OTC 2002 snowfree date flower open flower shed Figure 21 - Flowering phenology in V.Vitis-idaea in OTCs and control plots. Data points are means of 4 replicates for all phenological events in OTCs for both years. They are means of 4, 2 and 2 replicates for control 2001 and means of 20,15 and 15 replicates for control 2002 for the phenological events of snow-free date, flowers open and flowers shed, respectively. Error bars are ± 1 standard deviation. No data exists for snow-free date for OTCs in 2001. Asteriks indicate significant difference in overall means between control and OTCs. 69 5 4 3 2 1 0 -1 -2 a) n=20 a I number of flowers ) number of fruit n=4 n=4 n=16 2001 control OTC2D01 control OTC 2002 2002 jo 1.5 01 1 1 | 0.5 0 133=1 control OTC 2001 control OTC 2002 2001 2002 mean annual branch growth b) Figure 22 - Reproductive investment and success (a) and mean annual branch growth (b) in V.Vitis-idaea. Samples sizes for mean branch growth are 20 for controls and 4 for OTCs. Samples sizes for flower and fruit means are as indicated. There were no data for number of flowers in 2002 control plots. Error bars are ±1 standard deviation. Lower-case letters indicate significant differences. The analysis of plant responses in E. vaginatum (Table 25) showed that snow-free date was affected by the presence of OTCs; however, the effect depended on year. In 2001, snowmelt occurred 11 days later in the OTCs while in 2002, it was only one day later. Date of leaf emergence, pollen release and seed set were not influenced by simulated warming (Table 25). The prefloration phenophase was significantly shortened in OTCs (11 days in 2001 and one day in 2002) compared to control plots, with the amount affected by year (Figure 23). Although reproductive investment was not affected by simulated warming, growth was (Figure 24a). There was an overall decrease in leaf length in the OTCs as compared to controls and magnitude of response depended on year (Figure 24b). 70 Table 25 - Output from Analysis of Variance on plant responses in E. vaginatum from control plots and within OTCs. Year effects and the interaction between treatment and year were analysed when sufficient data were present. The variables leaf emergence, pollen release, seed set and floration were rank transformed prior to analysis. Leaf-out is number of days from snow-free date to leaf emergence. Prefloration is number of days from snow-free date to time of pollen release. Floration is time from pollen release to seed set. Variable Sample size Source F-value P-value Snow-free date 67 Treatment 42.29 0.0074 Year 0.26 0.6117 Treat*year 27.28 <0.0001 Leaf emergence 40 Treatment 5.20 0.1068 Pollen release 54 Treatment 1.02 0.3863 Year 5.54 0.0231 Treat*year 0.72 0.3393 Seed set 47 Treatment 2.11 0.2426 Year 26.99 <0.0001 Treat*year 0.85 0.3636 Leaf-out 40 Treatment 0.01 0.9174 Prefloration 44 Treatment 10.36 0.0486 Year 0.09 0.7611 Treat*year 6.04 0.0192 Floration 47 Treatment 0.16 0.7189 Year 66.68 <0.0001 Treat*year 0.63 0.4339 No. inflorescence stalks 40 Treatment 1.01 0.3899 Year 0.99 0.3275 Leaf length 80 Treatment 15.70 0.0287 Year 91.46 O.0001 Treat*year 4.09 ' 0.0468 71 220 140 -I , - , , snowfree date pollen release leaf-out seed set Figure 23 - Flowering phenology in E. vaginatumin OTCs and control plots. Data points are means of 4 replicates for all phenological events in OTCs for both years. They are means of 7,14 and 10 replicates in control 2001 for the phenological events of snow-free date, pollen release and seed set, respectively. No data exists for leaf-out in 2001. Data are means of 19 replicates for snow-free, pollen release, leaf-out and 10 replicates for seed set in control 2002 Error bars are + 1 standard deviation. Asteriks indicate significant difference in overall means, as well as a significant effect of year on treatment. 12 10 n=20 n=16 control -Otc2001 control _ otc 2002 2001 2002 mean annual number of inflorescences a) 300 250 200 150 100 50 0 b) control 2001 otc 2001 control 2002 otc 2002 mean annual leaf length Figure 24 - Reproductive effort (a) and mean leaf length (b) in E. vaginatum. Sample sizes for mean leaf length are 20 for controls and 4 for OTCs. Sample sizes for mean inflorescence number is as indicated. Error bars are ± 1 standard deviation. No data were available for number of inflorescences in control plots, 2002. Lower-case letters indicate significant differences. 7 2 3.4 Discussion The open-topped chambers increased air and soil temperatures overall, with more of an effect from noon to early evening and virtually no effect overnight. This pattern is consistent with other studies using similar types of chambers (e.g. Wookey etal. 1993, Chapin etal. 1995, Sandvik and Totland 2000). It has been argued that such a pattern does not adequately simulate a climate warming scenario for the Arctic and that there are other confounding effects such as: altered snow accumulation, humidity and dew formation; altered spectral distribution of incident radiation; inhibition of the diffusive mixing of carbon dioxide and other gases; the reduction of evapo-transpiratipn in the summer, and; the reduction of mechanical damage caused by ice crystal abrasion in the winter (Kennedy 1995). These types of effects were not monitored in this study; however, they have been examined by Marion et al. (1997) and Hollister and Webber (2000). Both reviews determined that open-topped greenhouse designs sufficiently increase temperatures while minimizing the undesirable ecological effects and therefore, are useful tools in experimental warming research in arctic ecosystems. In terms of plant responses to warming, the OTCs in the tussock tundra habitat delayed snowmelt by an average of six days within the chambers. Despite < this later snowmelt, leaf emergence, pollen release and seed set in E. vaginatum occurred at virtually the same time in the OTCs as in control plots. Essentially, the prefloration phenophase was contracted in this species to allow flowering to occur at the same time while floration time remained unchanged. A similar response to simulated warming was observed for one of the study species in the lichen heath 73 habitat (V.Vitis-idaea). Again, prefloration was contracted whereas floration time did not change. The other study species in the lichen heath habitat, L. decumbens, responded to warming by having flowers open earlier and shed earlier with no effect on prefloration phase length. The response of L. decumbens seems counterintuitive; if snowmelt occurred at the same time in both treatments but flowering occurred earlier in warmed plots how could the prefloration phase length not differ? There were no data for snow-free date for OTCs in 2001, but flowering occurred obviously earlier in the OTCs in this year. There was likely not enough power in the analysis of snow-free date to determine any difference. The lack of snow-free data for 2001 also precluded the use of prefloration phase length for that year. The results of this study confirm results extant in the literature; that prefloration phase is contracted in arctic plants that are experimentally warmed (Alatalo and Totland 1997, Jones etal. 1997, Levesque etal. 1997, Molau 1997a, Molau and Shaver 1997, Stenstrom and Jonsdottir 1997, Welker et al. 1997, Arft et al. 1999). There is also support in the literature for a more pronounced effect of snowmelt date on earlier phenological events (i.e. prefloration) rather than later (i.e. floration) (e.g. Johnstone 1995). There are a few examples where species were not phenologically responsive to experimental warming (Stenstrom ef al. 1997 and Totland 1999). Both species in these examples, S. oppositifolia and Ranunculus acris, flower very early in the season under natural conditions and perhaps demonstrate that there are physiological limits to prefloration phase length that preclude any further contraction. Given that L. decumbens is not an early flowering 74 species it is not likely that this explanation holds, and this further substantiates the possibility of low power in the analysis. The three study species showed varied growth responses to warming. Branch growth in L. decumbens was greater in warmed plots than controls whereas, leaf length in E. vaginatum was less in warmed plots. V. Vitis-idaea showed a mixed response with branch growth increasing in the first year of experimental warming and decreasing in the second year. These responses are consistent with some published results; however, there are inconsistent results in the literature. Chapin et al. (1995) reported no short-term response of evergreen dwarf shrubs (including V.Vitis-idaea and L. decumbens) to simulated warming but did find a positive growth response of E. vaginatum. These results were substantiated in a meta-analysis of 13 experiments across the circumarctic, which demonstrated that dwarf evergreen shrubs showed no growth response to temperature enhancement while graminoids showed a slight positive response (Arft et al. 1999). In contrast, Parsons ef al. (1994) found V.Vitis-idaea to respond positively to increased temperature in terms of stem length, mass and number of leaves after two years of experimentation. Similarly, Ledum palustre increased shoot production and aboveground biomass accumulation under enhanced temperature in a Japanese alpine fellfield (Kudo and Suzuki 2003). In another meta-analysis of 36 experiments on effects of warming and other environmental manipulations on arctic plants, evergreen shrubs increased biomass production and sedges showed no change (Dormann and Woodin 2002). 75 The positive growth responses found in the current study for L. decumbens and V.Vitis-idaea (in the first year) confirm the results reported by Parsons ef al. (1994), Dormann and Woodin (2002) and Kudo and Suzuki (2003) and contradict those of Chapin ef al. (1995) and Arft ef al. (1999). The negative growth response of E. vaginatum to warming was surprising but similar to results from Hobbie ef al. (1999) where aboveground biomass production decreased in warmed plots. The mechanism for this response is unclear, but could be related to nutrient limitation. Shaver et al. (1986) reported that E. vaginatum stopped growth when nutrient concentration in its tissues were reduced to a minimum concentration. Further, Rolph (2003) found that soil microbes immobilized nitrogen in experimentally warmed plots situated in wet habitats in the Canadian High Arctic. The negative response of leaf growth, however, could also have been compensated by other methods of production that were not measured such as increased tillering and root growth (Shaver etal. 1986). V.Vitis-idaea showed a decrease in flower production in warmed plots as compared to the control which is counter to what has been reported elsewhere. For example, Kudo and Suzuki (2003) found no change in flower production in experimentally warmed V.Vitis-idaea in an alpine location after five years of warming. £ vaginatum showed no difference in number of inflorescences in the current study. And although ovule number per inflorescence and seed weight have been shown to increase with warming (Molau and Shaver 1997), no comparable measure of reproductive effort under conditions of experimental warming for E. vaginatum was found in the literature. 76 Both species in the lichen heath habitat showed ho response in reproductive success (i.e. fruit production) to simulated warming. This is consistent with Wookey et al. (1993) who found no reproductive response to increased temperature in another evergreen dwarf shrub, Empetrum hermaphroditum, in a sub-Arctic shrub heath site. However, this contradicts that reported by Dormann and Woodin (2002) in their meta-analysis, where all plant functional types showed an increase in reproduction with experimental warming. It has been suggested that low arctic systems might be more inclined to show vegetative responses to warming (Wookey et al. 1993); or that initial responses to warming are vegetative followed by reproductive responses after a number of years (Arft et al. 1999). The three study species all showed growth responses (albeit one in the negative) and therefore seem to support the notion of initial vegetative responses to simulated warming, in the Low Arctic. Reproductive effort and success showed little response to warming over the two years of the experiment. Because of the short-term nature of the current study, a shift to reproductive responses may not be expected at this stage. Interestingly, reproductive effort was negatively affected in one species in the first year, perhaps as a result of the high investment in growth. This may be evidence for resource allocation trade-offs that have been observed for other tundra species (e.g. Johnstone and Henry 1997, Wyka and Galen 2000). 3.5 Conclusions Under warmer temperatures, the spring phenology of three low arctic plant species was advanced such that the time from snowmelt to flowering contracted. Timing of the end of the flowering phase remained unchanged, however. Growth in 77 two evergreen dwarf shrub species increased when experimentally warmed and growth of one species of sedge was decreased. Reproductive effort and success were essentially unchanged over the two-year experiment except for V. Vitis-idaea in the first year of warming when number of flowers was reduced. Results from this study complement the literature. Further long-term experimentation will be essential to determine whether the observed growth response is an initial response that will be replaced by a later reproductive response. It is perhaps pertinent to note that short-term studies often do not reflect longer-term plant responses in arctic ecosystems (e.g. Chapin etal. 1995) and, therefore, long-term projects studying potential impacts of climate warming are essential to gain an accurate depiction of vegetation change over time. 78 C H A P T E R IV - CONCLUSIONS AND RECOMMENDATIONS FOR F U R T H E R R E S E A R C H 4.1 Introduction Previous chapters presented alternative techniques used to examine plant interactions with their climatic environment. Canonical Correlation Analysis was used in Chapter II, to form relationships between plant responses (phenology, growth and reproduction) and a suite of environmental and previous investment variables using observational data. In Chapter III, three of the eight study plants were shown to have responded to experimental warming over two years in an experimental study. This final chapter will assess how well results from the two analyses complement each other and what further studies can enhance those already underway at the Tundra Ecosystem Research Station. 4.2 Synthesis The results of the experimental warming of L. decumbens are somewhat consistent with the relationships found through the analysis of the observational data. Prefloration time in L. decumbens did not respond to temperature in the OTC manipulation consistent with the finding from the C A N C O R analysis that prefloration was not well predicted by temperature. Conversely, floration time was also not influenced by increased temperatures in the OTCs yet from the C A N C O R analysis was negatively related to temperature (higher temperatures, shorter floration). This may simply be due to lack of data in 2001 as discussed in section 3.3. 79 Both prefloration and floration phase length in V.Vitis-idaea were negatively related to temperature in the C A N C O R analysis yet in the experimental manipulation, only prefloration time was significantly shortened by simulated warming in the OTCs. Numbers of flowers appeared not to be influenced by temperature in the C A N C O R analysis yet were reduced in the first year of warming. Number of fruit appeared to be moderately related to temperature in a positive manner yet simulated warming did not increase fruit production. Results from the experimental warming of E. vaginatum are partially consistent with the relationships revealed in the C A N C O R analysis. In confirmation of the C A N C O R analysis, which found prefloration to be negatively correlated with temperature, prefloration time was shortened in experimentally warmed plants. In contrast, leaf length was reduced with simulated warming, yet the C A N C O R analysis showed a positive relationship between temperature and leaf length in this species. In general, the phenological results of the experimental manipulation confirmed those from the C A N C O R analysis of observational data except in the case of floration intervals. Growth and reproductive responses to warming, however, were not consistent to those of the C A N C O R analysis. Given that the C A N C O R analysis utilized six years of data and the experimental warming component only incorporated two years, the former relationships are likely more reliable. Further, because the C A N C O R analysis represents average responses over time, small-scale variability will be diminished in favour of longer-term trends. However, it is clear that experimental and observational studies complement each other and both are needed to elucidate plant responses. 80 4.3 Recommendations for further work Other studies have reported that plant responses to simulated warming are often constrained by nutrient availability (Shaver etal. 1986, Chapin and Shaver 1996), competition (Chapin et al. 1995) and/or internal patterns of resource allocation (Johnstone and Henry 1997). In light of these constraints, the work at the Tundra Ecosystem Research Station would be enhanced by the initiation of research that addressed these factors. Examining soil nutrient fluxes inside and outside OTCs would provide information on how nutrient pools in the soil are changing over time with and without simulated warming. Further, community composition work would aid in establishing how individual species responses fit into the larger picture of competition and dominance in the low arctic vegetation community. What has become clear after these two studies and the work of others is that short-term studies are not sufficient to determine plant responses to climate and climate warming. Arctic plants display a lag in reproductive responses to climate as reproductive structures are developed one to several years before opening. Further, initial growth responses are often constrained by nutrient limitations, competitive interactions and/or internal patterns of resource allocation. Within this context, long-term ecological monitoring of tundra ecosystems is a necessity. 81 R E F E R E N C E S Arft, A .M. , M.D. Walker, J . Gurevitch, J . M. Alatalo, M.S. Bret-Harte, M. Dale, M. Diemer, F. Gugerli, G.H.R. Henry, M.H. Jones, R.D. Hollister, I.S. Jonsdottir, K. Laine, E. Levesque, G.M. Marion, U. Molau, P. Molgaard, U. Nordenhall, V. Raszhivin, C. H. Robinson, G. Starr, A. Stenstrom, M. Stenstrom, O. Totland, P.L. Turner, L. J . Walker, P.J. Webber, J . M. Welder, and P.A. Wookey. 1999. Responses of tundra plants to experimental warming: meta-analysis of the International Tundra Experiment. Ecological Monographs 69: 491-511. Alatalo, J .M. and O. Totland. 1997. Response to simulated climatic change in an alpine and subarctic pollen-risk strategist, Silene acaulis. Global Change Biology 3 (Suppl. 1): 74-79. Bean, D. and G.H.R. Henry. 2002. Climate change impacts on tundra ecosystems: the C A N T T E X network of ecological monitoring sites in the Canadian Arctic. Unpublished manuscript. University of British Columbia, Canada. Bliss, L.C. 1956. A comparison of plant development in microenvironments of arctic and alpine tundra. Ecological Monographs 26: 303-337. Bliss, L.C. 1971. Arctic and alpine life cycles. Annual Review of Ecology and Systematics 2: 405-438. Chapin, F.S.III. and G. Shaver. 1985. Individualistic growth response of tundra plant species to environmental manipulations in the field. Ecology 66: 564-576. Chapin, F.S. Ill and G.R. Shaver. 1996. Physiological and growth responses of arctic plants to a field experiment simulating climatic change. Ecology 77: 822-840. Chapin F. Ill, G. Shaver, A. Giblin, K. Nadelhoffer and J . Laundre. 1995. Responses of arctic tundra to experimental and observed changes in climate. Ecology 76: 694-711. Dillon, W.R. and M. Goldstein. 1984. Multivariate Analysis: Methods and Applications. John Wiley & Sons, New York. Dormann, C. and S. Woodin. 2002. Climate change in the Arctic: using plant functional types in a meta-analysis of field experiments. Functional Ecology 16:4-17. Elzinga, C , D. Salzer and J . Willoughby. 1998. Measuring and Monitoring Plant Populations. Bureau of Land Management, Denver, Colorado. Fetcher, N. and G.R. Shaver. 1990. Environmental sensitivity of ecotypes as a potential influence on primary productivity. American Naturalist 136: 126-131. 82 Fitter, A .H. , R.S.R. Fitter, L.T.B. Harris and M. H. Williamson. 1995. Relationships between first flowering date and temperature in the flora of a locality in central England. Functional Ecology 9: 55-60. Galen, C. and M.L. Stanton. 1991. Consequences of emergence phenology for reproductive success in Ranunculus adoneus (Ranunculaceae). American Journal of Botany 78: 978-988. Galen, C. and M.L. Stanton. 1995. Responses of snowbed plant species to changes in growing-season length. Ecology 76: 1546-1557. Hobbie, S., A. Shevtsova and F. Chapin. 1999. Plant responses to species removal and experimental warming in Alaskan tussock tundra. Oikos 84: 417-434. Hollister, R. and P. Webber. 2000. Biotic validation of small open-topped chambers in a tundra ecosystem. Global Change Biology 6: 835-842. Hurlbert, S. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54: 187-211. Inouye, D.W. and A.D. McGuire. 1991. Effects of snow-pack on timing and abundance of flowering in Delphinium nelsonii (Ranunculaceae): Implications for climate change. American Journal of Botany 78: 997-1001. Inouye, D.W., M.A. Morales and G.J . Dodge. 2002. Variation in timing and abundance of flowering by Delphinium barbeyi Huth (Ranunculaceae): the roles of snowpack, frost, and La Nina, in the context of climate change. Oecologia 130: 543-550. Johnstone, J . 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. Thesis. University of British Columbia, Vancouver, Canada. Johnstone, J . and G.H.R. Henry. 1997. Retrospective analysis of growth and reproduction in Cassiope tetragona and relations to climate in the Canadian High Arctic. Arctic and Alpine Research 29: 459-469. Johnstone, J . , U. Molau and G. Marion. 1996. Experimental design and statistical analysis in ITEX. Pages 46-48 in U. Molau and P. Molgaard, editors. International Tundra Experiment Manual, 2 n d edition. Danish Polar Center. Jones, M.H., C. Bay and U. Nordenhall. 1997. Effects 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): 55-60. 83 Kennedy, A.1995. Simulated climate change: are passive greenhouses a valid microcosm for testing the biological effects of environmental perturbations? Global Change Biology 1: 29-42. Kudo, G . 1991. Effects of snow-free period on the phenology of alpine plants inhabiting snow patches. Arctic and Alpine Research 23: 436-443. Kudo, G. and S. Suzuki. 2003. Warming effects on growth, production and vegetation structure of alpine shrubs: a five-year experiment in northern Japan. Oecologia 135: 280-287. Levesque, E., G.H.R. Henry and J . Svoboda.1997. Phenological and growth responses of Papaver radicatum along altitudinal gradients in the Canadian High Arctic. Global Change Biology 3 (Suppl. 1): 125-145. Marion, G.M. 1996. Temperature enhancement experiments. Pages 17-22 in U. Molau and P. Molgaard, editors. International Tundra Experiment Manual, 2 n d edition. Danish Polar Center. Marion, G. ,G. Henry, D. Freckmand, J . Johnstone, G. Jones, M. Jones, E. Levesque, U. Molau, P. Molgaard, A. Parsons, J . Svoboda and R. Virginia. 1997. Open-top designs for manipulating field temperature in high-latitude ecosystems. Global Change Biology 3 (Suppl. 1): 20-32. Marsden, R. 1992. Snowmelt, flowering phenology and reproductive success in tundra plants. Thesis. University College of North Wales, Bangor. Matthews, G .J . and R. Morrow Jr. 1985. CANADA and the World: An Atlas Resource. Prentice-Hall Canada Inc., Scarborough, Ontario. Matthews.S., H. Epp and G. Smith. 2001. Vegetation Classification for the West Kitikmeot/Slave Study Region. Final Report to the West Kitikmeot/Slave Study Society. Northwest Territories, Canada. Maxwell, B. 1992. Arctic climate: potential for change under global warming. Pages 11-34 in F.S. Chapin III, etal., editors. Arctic Ecosystems in a Changing Climate: An Ecophysiological Perspective. Academic Press, Inc., San Diego, California. Maxwell, B. 1997. Recent climate patterns in the Arctic. Pages 21-46 in W .C . Oechel et. al., editors. Ecological Studies 124: Global change and Arctic terrestrial ecosystems. Springer-Verlag, Berlin, Germany. Molau, U. 1993. Relationships between flowering phenology and life history strategies in tundra plants. Arctic and Alpine Research 25: 391-402. 84 Molau, U. 1996. Climatic impacts on flowering, growth and vigour in an arctic-alpine cushion plant, Diapensia lapponica, under different snow cover regimes. Ecological Bulletins 45: 210-219. Molau, U. 1997a. Responses to natural climatic variation and experimental warming in two tundra plant species with contrasting life forms: Cassiope tetragona and Ranunculus nivalis. Global Change Biology 3 (Suppl. 1): 97-107. Molau, U. 1997b. Phenology and reproductive success in Arctic plants: susceptibility to climate change. Pages 153-170 in W .C . Oechel etal., editors. Ecological Studies 124: Global change and Arctic terrestrial ecosystems. Springer-Verlag, Berlin, Germany. Molau,U. and S. Edlund. 1996. Plant response variables. Pages 23-38 in U. Molau and P. M0lgaard, editors. International Tundra Experiment Manual, 2 n d Edition. Danish Polar Center. Molau, U. and G. Shaver. 1997. Controls on seed production and seed germinability in Eriophorum vaginatum. Global Change Biology 3 (Suppl. 1): 80-88. Parsons, A. , J . Welker, P. Wookey, M. Press, T. Callaghan and J . Lee. 1994. Growth responses of four sub-Arctic dwarf shrubs to simulated environmental change. Journal of Ecology 82: 307-318. Pop, E.W., S.R. Oberbauer and G. Starr. 2000. Predicting vegetative bud break in two arctic deciduous shrub species, Salix pulchra and Betula nana. Oecologia 124: 176-184. Rathcke, B. and E.P. Lacey. 1985. Phenological patterns of terrestrial plants. Annual Review of Ecology and Systematics 16: 179-214. Rolph, S .G . 2003. Effects of a ten-year climate warming experiment on nitrogen cycling in high arctic tundra. Thesis. University of British Columbia, Vancouver, Canada. Sandvik, S.M. and O. Totland. 2000. Short-term effects of simulated environmental changes on phenology, reproduction, and growth in the late-flowering snowbed herb Saxifraga stellaris L. Ecoscience 7: 201-213. Shaver, G. , F. Chapin HI, B. Gartner. 1986. Factors limiting seasonal growth and peak biomass accumulation in Eriophorum vaginatum in Alaskan tussock tundra. Journal of Ecology 74: 257-278. Sorensen, T. 1941. Temperature relations and phenology of the northeast Greenland flowering plants. Meddelelser om Granland 125: 1-305. 85 Sparks, T.H., E.P. Jeffree and C E . Jeffree. 2000. An examination of the relationship between flowering times and temperature at the national scale using long-term phenological records from the UK. International Journal of Biometeorology 44: 82-87 Starr, G. , S.F. Oberbauer and E.W. Pop. 2000. Effects of lengthened growing season and soil warming on the phenology and physiology of Poygonum bistorta. Global Change Biology 6: 357-369. Stenstrom, A. and I.S. Jonsdottir. 1997. Responses of the clonal sedge, Carex bigelowiix to two seasons of simulated climate change. Global Change Biology 3 (Suppl. 1): 89-96. Stenstrom, ML, F. Gugerli and G.H.R. Henry. 1997. Response of Saxifraga oppositifolia L. to simulated climate change at three contrasting latitudes. Global Change Biology 3 (Suppl. 1): 44-54. Tabachnick, B.G. and L.S Fidell. 1996. Using Multivariate statistics, 3 r d edition. Harper Collins College Publishers, New York. ter Braak, C.J .F. 1995. Ordination. Pages 91-173 in R.H.G. Jongman, C.J .F. ter Braak and O.F.R. van Tongeren, editors. Data Analysis in Community and Landscape Ecology. Cambridge University Press, Cambridge. Thorhallsdottir, T.E. 1998. Flowering phenology in the central highland of Iceland and implications for climatic warming in the Arctic. Oecologia 114: 43-49. Totland, 0 .1999 . Effects of temperature on performance and phenotypic selection on plant traits in alpine Ranunculus acris. Oecologia 120: 242-251. Walther, G. , E. Post, P. Convey, A. Menzel, C. Parmesan, R. Beebee, J . Fromentin, O. Hoegh-Guldberg and F. Bairlein. 2002. Ecological responses to recent climate change. Nature 416:389-395. Walker, M.D., R.C. Ingersoll and P.J . Webber. 1995. Effects of interannual climate variation on phenology and growth of two alpine forbs. Ecology 76: 1067-1083. Welker, J .M. , U. Molau, A .N . Parsons, C H . Robinson and P.A. Wookey.1997. Response of Dryas octopetala to ITEX environmental manipulations: a synthesis with circumpolar comparisons. Global Change Biology 3 (Suppl. 1): 61-73. Woodley, E.J. and Svoboda, J . 1994. Effects of habitat on variations of phenology and nutrient concentration among four common plant species of Alexandra Fiord Lowland. Pages 157-176 in J . Svoboda and B. Freedman, editors. Ecology of a polar desert, Alexandra Fiord, Ellesmere Island, Canada. Captus University Publications, Toronto. 86 Wookey, P., A. Parsons, J . Welker, J . Potter, T. Callaghan, J . Lee and M. Press. 1993. Comparative responses of phenology and reproductive development to simulated environmental change in sub-arctic and high arctic plants. Oikos 67: 490-502. Wyka, T and C. Galen. 2000. Current and future costs of reproduction Oxytropis sericea, a perennial plant from the Colorado Rocky Mountains, USA. Arctic Antarctic and Alpine Research 32: 438-448. Zimmerman, M. and R. Gross. 1984. The relationship between flowering phenology and seed set in an herbaceous perennial plant, Polensonium foliosissimum Gray. American Midland Naturalist 111:185-191. 87 

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.831.1-0091621/manifest

Comment

Related Items