UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The relation between climate and abundance cycles in barren-ground caribou herds of the Northwest Territories,… Zalatan, Rebecca 2006

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

Item Metadata

Download

Media
831-ubc_2006-200261.pdf [ 13.96MB ]
Metadata
JSON: 831-1.0092761.json
JSON-LD: 831-1.0092761-ld.json
RDF/XML (Pretty): 831-1.0092761-rdf.xml
RDF/JSON: 831-1.0092761-rdf.json
Turtle: 831-1.0092761-turtle.txt
N-Triples: 831-1.0092761-rdf-ntriples.txt
Original Record: 831-1.0092761-source.json
Full Text
831-1.0092761-fulltext.txt
Citation
831-1.0092761.ris

Full Text

THE RELATION BETWEEN CLIMATE AND ABUNDANCE CYCLES IN BARREN-GROUND CARIBOU HERDS OF THE NORTHWEST TERRITORIES, CANADA. by REBECCA ZALATAN B.A., University of Ottawa, 2000 M.Sc., University of Ottawa, 2002 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Geography) THE UNIVERSITY OF BRITISH COLUMBIA September 2006 © Rebecca Zalatan, 2006 A B S T R A C T The central objective of this research was to determine i f there is a relationship between long-term barren-ground caribou (Rangifer tarandus) abundance and climate patterns in the Canadian Arctic. A long-term dataset indexing caribou abundance was obtained from the frequency o f trampling scars on tree roots o f black spruce (Picea mariana [Mi l l . ] BSP) in the forest-tundra of central Northwest Territories. Samples were collected from roots of live trees along well-used migration trails in the forest tundra. Two groups o f sites were sampled that roughly corresponded with the late summer migration routes o f the Bathurst and Beverly caribou herds. The scar frequency distributions were dated from A . D . 1760-2000 and both groups o f sites showed similar abundance patterns through time. To best determine the relation between long-term climate and the proxy caribou abundance, local climate data were needed. However, the only annually-resolved climate data in the region is the short length o f record (63 years) available from Yellowknife (which is up to 300 km away from the furthest sites in this study). Therefore, I developed a series of tree-ring chronologies at seven o f the nineteen sites where caribou abundance was reconstructed, in an effort to increase the knowledge o f climate variability in this region. Summer temperatures (July-August) were reconstructed using standard dendroclimatological techniques. The final objective of my research was to determine i f there was a correlation between patterns in the summer or winter index of the Arctic Oscillation (AO) and long-term caribou abundance. I determined that the A O was most closely correlated to summer temperatures (June-August) at Yellowknife. The summer index o f the A O has undergone four major phase changes during the n last century. Wavelet coherence demonstrated that the relation between caribou abundance cycles and the A O S changed from inversely related during the first two phases of the A O S , to in-phase during the final two phases of the A O S . This study was the first to demonstrate the complexities associated with relating long-term trends in the A O to abundance cycles of barren-ground caribou. Additionally, this study was the first to illustrate the importance of obtaining long-term datasets when relating large-scale climate to caribou abundance cycles. 111 T A B L E O F C O N T E N T S ABSTRACT II T A B L E OF CONTENTS IV LIST OF TABLES V LIST OF FIGURES VI ACKNOWLEDGEMENTS X CHAPTER 1. INTRODUCTION 1 1.1 OBJECTIVES '. 8 1.2 THESIS OUTLINE 9 1.3 REFERENCES 10 CHAPTER 2. LONG-TERM ABUNDANCE PATTERNS OF BARREN-GROUND CARIBOU USING TRAMPLING SCARS ON ROOTS OF PICEA MARIANA IN THE NORTHWEST TERRITORIES 16 2.1 INTRODUCTION 16 2.2 M E T H O D S 18 2.3 RESULTS 28 2.4 DISCUSSION 33 2.5 REFERENCES •. 37 CHAPTER 3. SPATIO-TEMPORAL VARIATION OF TREE-RING GROWTH A L O N G A TRANSECT AT TREELINE WITHIN T H E RANGE OF BARREN-GROUND CARIBOU: A DENDROCLIMATIC APPROACH : 39 3.1 INTRODUCTION 39 3.2 M E T H O D S : 42 3.3 RESULTS 49 3.4 DISCUSSION 62 3.5 REFERENCES : .- 65 CHAPTER 4. THE RELATION BETWEEN LONG-TERM ABUNDANCE CYCLES IN BARREN-GROUND CARIBOU AND T H E ARCTIC OSCILLATION 71 4.1 INTRODUCTION .-. .' 71 4.2 M E T H O D S . , 74 4.3 RESULTS : 78 4.4 DISCUSSION 90 4.5 REFERENCES '. 96 CHAPTER 5. THESIS SUMMARY AND SYNTHESIS 103 APPENDIX A 106 iv LIST OF TABLES TABLE 2.1 Geographic location, number of scars and samples for the northwest and southeast sites, Northwest Territories. Locations are also shown in Fig. 2.1 28 TABLE 3.1 Site characteristics of Picea mariana chronologies along treeline in the N.W.T 43 TABLE 3.2 Descriptive statistics for the seven chronologies developed using Picea mariana. s.d. = standard deviation; r l = first-order autocorrelation; SNR = signal-to-noise ratio; V A R p c l = variance explained by the first component; EPS = expressed population signal 51 TABLE 3.3 Pearson's correlations among the seven residual chronologies developed from samples taken at treeline, N.W.T. A l l correlations were significant at p<0.05; «=88. 52 TABLE 3.4 Pearson's correlations among the seven residual treeline chronologies and the ITRDB chronologies (Aus. = Austin; Copp. = Coppermine; Mac. = Mac Kinley; Pet. = Pethai.). Marked correlations are significant at p<0.05, «=62 53 TABLE 3.5 Calibration and verification statistics for predicting July-August temperatures from eigenvectors derived from five chronologies at treeline in the N.W.T 59 TABLE A. l Spearman's rank correlations between scar frequency distribution for both groups of sites (northwest and southeast) annually, in 2-year age-classes, 3-year age-classes and 4-year age-classes. Correlations are significant at p < 0.05 106 TABLE A.2 Pearson's correlations among the seven tree-ring chronologies and mean monthly temperature. A l l marked correlations are significant atp<0.05; n = 59 109 TABLE A.3 Pearson's correlations among the seven tree-ring chronologies and total monthly precipitation. A l l marked correlations are significant at j?<0.05; n = 59 110 TABLE A.4 Response function analysis between the monthly index of the A O and the seven tree-ring chronologies over the entire study period from May of the previous growth year to August of the current growth year. Marked correlations are significant at p <0.05 I l l TABLE A.5 Spearman's rank correlation between the July index of the A O and the seven chronologies for each phase. Marked correlations are significant at p <0.05 I l l LIST OF FIGURES FIGURE 1.1 The seasonal range and calving grounds of some barren-ground caribou herds in the Northwest Territories and Nunavut 2 FIGURE 1.2 Sea-level pressure and associated temperature anomalies as related to the A O from December-February. The A O is a major mode of circulation variability in the Arctic 6 FIGURE 1.3 A conceptual model linking the phases of the Arct ic Oscillation, winter and summer climate conditions, and caribou abundance 8 FIGURE 2.1 Sampling sites of scarred roots on Picea mariana (sampled 2002) and the summer migratory route of barren-ground caribou. The geographic coordinates o f the sampling locations for the scarred roots are listed in Table 2.1 19 FIGURE 2.2 This is an example of a typical site where caribou scars were collected. This photo was taken from a helicopter (at approximately 300 m elevation) and is located in the forest-tundra within the annual range of the Bathurst caribou herd 20 FIGURE 2.3 Caribou migration trails across an esker taken from a helicopter in the central N . W . T 21 FIGURE 2.4 Bathurst caribou migrating across the landscape in the N . W . T 21 FIGURE 2.5 A series o f black spruce roots trampled by barren-ground caribou. Trampling removes part of the bark, leaving behind a scar which is used to represent caribou abundance 22 FIGURE 2.6 Tracks of 10-20 satellite-collared caribou cows from the Bathurst herd (1996-2002). The black boxes outline the general area where the study sites are located 23 FIGURE 2.7 Schematic diagram of a scar on a cross-sectional sample of a root 24 FIGURE 2.8 Scar frequency distribution (a) using all samples from sites 1-15 (blue bars) and sites 16-19 (grey bars) and cumulative frequencies of the number of samples (lines) (r s = 0.88; /?<0.05, n - 24); and (b) using only scars on roots established before 1900 (r s = 0.63, p<0.05, n = 24) 30 vi FIGURE 2.9 Bathurst herd caribou numbers from aerial photography surveys (1984-2003; black line; means ± SE), and abundance patterns (high = dark grey shading or low = light grey shading) derived from traditional knowledge o f elders from the Dogrib First Nation relative to the scar frequency distribution for the northwest sites (Bathurst herd; bars) 32 FIGURE 2.10 Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups of sites 33 FIGURE 3.1 Map showing location of tree-ring chronology sites (numbered sites), sites from the International Tree-Ring Databank ( ITRDB) , Yellowknife meteorological station in the N . W . T . and the summer migratory route of barren-ground caribou 43 FIGURE 3.2 The standardized Arct ic Oscillation (AO) anomaly index for June to August for the period between 1900-2002, and a 5 t h order regression (data from http://www.jisao.washington.edU/ao/#monthly). The values of the A O S have been multiplied by . -1 to illustrate the positive and negative phases 49 FIGURE 3.3 Ring-width chronologies from the seven sampled sites (ring-width sample depth is below each chronology curve). The thick line is a five-year moving average 50 FIGURE 3.4 Hierarchical cluster analysis of seven residual chronologies 54 FIGURE 3.5 Plot of the loadings for the first two principal components for each chronology. The percentage of explained variance for P C I and PC2 is 61.80% and 13.10%, respectively...54 FIGURE 3.6 Percent variance expressed by the first and second principal components of P C A derived from the seven residual chronologies for the subperiods: 1916-1946, 1926-1956, 1936-1966, 1946-1976, 1956-1986, and 1966-1996. For the whole common period 1916-2003, the variance explained by the first and second principal components is 74.9% 55 FIGURE 3.7 Response function coefficients relating mean monthly temperature and total monthly precipitation to ring-width chronologies at treeline sites, N . W . T . Shaded coefficients are significant at p<0.05 56-57 FIGURE 3.8 Observed (dashed line) and reconstructed (solid line) July-August temperatures at treeline, N . W . T . Darker solid line is a 3-year running mean of the reconstructed July-August temperatures. Reconstruction covers the time span 1831-2002 60 FIGURE 3.9 The standardized Arctic Oscillation (AO) anomaly index for June to August for the period between 1900-2002, and a 5 t h order regression curve (data from http://tao.atmos.washington.edu/ao/#monthlv). The values o f the A O S have been multiplied by -1 to illustrate the positive and negative phases 61 FIGURE 4.1 (A) Relationship between the summer index of the standardized Arct ic Oscillation (AOS) and mean monthly summer temperatures for the period 1943-2002; (B) Relationship vn between the winter index of the standardized Arctic Oscillation ( A O W ) and mean monthly summer temperatures for the period 1943-2001 78 FIGURE 4.2 The standardized Arctic Oscillation (AO) anomaly index for June to August and mean monthly temperatures for June to August for the comparative period of 1943-2001 79 FIGURE 4.3 Percent variance expressed by the first and second principal components of P C A based on a matrix with the monthly index of the Arctic Oscillation for the subperiods: 1900-1930, 1910-1940, 1920-1950, 1930-1960, 1940-1970, 1950-1980, 1960-1990, and 1970-2000. For the whole period 1900-2000, the variance explained by the first and second principal components is 50.3% 81 FIGURE 4.4 Standardized distribution of the number o f trampling scars (residuals of the log-linear regression) from 1900-2000 for the northwest and southeast sites 81 FIGURE 4.5 Residuals of the log-linear regression on the frequency distribution of trampling scars for both sets of sites grouped into five-year age-classes and a 5 t h order regression of the summer index o f the Arctic Oscillation (June-August) for the period 1900-2002 82 FIGURE 4.6 The standardized Arctic Oscillation (AO) anomaly index for October to M a y for the period between 1900-2002, (data from http://tao.atmos.washington.edu/ao/#monthly) 83 FIGURE 4.7 The cross wavelet transform (CWT) of the northwest caribou abundance (top panel) and the A O (bottom panel) 85 FIGURE 4.8 The cross wavelet transform (CWT) of the southeast caribou abundance (top panel) and the A O (bottom panel) 86 FIGURE 4.9 The cross wavelet transform (CWT) of the northwest caribou abundance (top panel) and the southeast caribou abundance (bottom panel) 86 FIGURE 4.10 The wavelet coherence (WTC) between the southeast caribou abundance and the northwest caribou abundance time series 87 FIGURE 4.11 The wavelet coherence (WTC) between the northwest caribou abundance and the A O time series 88 FIGURE 4.12 The wavelet coherence (WTC) between the southeast caribou abundance and the A O time series 89 FIGURE A . l Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups of sites in 1-year age-classes 107 FIGURE A.2 Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups of sites in 2-year age-classes 107 Vlll F I G U R E A.3 Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups of sites in 3-year age-classes 108 F I G U R E A . 4 Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups of sites in 4-year age-classes 108 ix A C K N O W L E D G E M E N T S I would like to thank my supervisor Dr. Greg Henry, whose guidance taught me to become an independent thinker. Greg always supported my decisions and allowed me to be where I needed to be to complete my dissertation. He always encouraged me to publish my work early and to attend conferences to expand my knowledge. I am indebted to my committee: Lor i Daniels, Anne Gunn, and Konrad Gajewski. Lor i was my sole link to the dendro community and she always encouraged me in a supportive way. Anne was an amazing person to have around, not just on my committee but in the field and around the office in Yellowknife. Konrad has been a constant source of inspiration and support during my Ph.D. I thank him for allowing me to work in his lab while I was l iving in Ottawa. His energy and enthusiasm did not go unnoticed. Also , I would like to acknowledge the support of the L P C for letting me take up space in their lab, and for all the great "fogs" that made my time there more enjoyable. I thank Dr . Shelly Rayback for her constant advice and support from the beginning of my Ph.D., right until the very end. She was an incredible source of help and was always encouraging and supportive. M y experience in the north would not have been the same without my host Kar in Clark and her loving family. Thank you so much for taking me in and showing me the beauty and spirit o f the north. I would like to thank J. Wil l iams, D . Abernethy, G . Furniss, D . Cluff, J. Kodzin , J. Mackenzie, P. Liske, J. Lee, and D . Johnson for field assistance and data collection. A support that cannot go unnoticed is my husband and best friend Martin Turpin. He was always reminding me to make schedules and write down my thoughts. Encouraging me to become a better person also made me a better academic. Thank you for standing by me until the very end. This research was financially supported by a Natural Sciences and Engineering Research Council of Canada ( N S E R C ) Scholarship, N S E R C Northern Supplement, The Association of Canadian Universities for Northern Studies Caribou Award ( A C U N S ) , Royal Canadian Geographical Society (RCGS) , and a U B C University Graduate Scholarship to R. Zalatan. Fieldwork was aided by a grant from the Northern Student Training Program (NSTP) to R. Zalatan, an N S E R C Discovery Grant and Northern Supplement to G .H.R . Henry and by the Department of Environment and Natural Resources, G N W T , Yellowknife. I acknowledge the logistic support of the Polar Continental Shelf Project (PCSP project N o 636-04)). x C H A P T E R 1. I N T R O D U C T I O N Caribou abundance across a landscape fluctuates from annual to decadal time scales (Skoog 1968). Determining changes in barren-ground caribou (Rangifer tarandus groenlandicus) abundance is limited by the practicalities o f counting large numbers o f animals whose movements and distribution are often unpredictable over thousands o f square kilometers (Messier et al. 1988). Barren-ground caribou are a migratory continental tundra subspecies that migrate for long distances from the forest to the open tundra (Mallory and Hi l l i s 1998). The seasonal range o f some barren-ground caribou herds is located throughout portions o f the central Northwest Territories and Nunavut (Fig. 1.1). Arguably, the most important stop in this arduous trek is the traditional calving grounds where the cows o f each o f the northern herds gather each year to bear their young. This territory makes up a small fraction o f their annual range, but they return to it each spring with some variation in exact timing or location from year to year (Gunn and Fournier 2000). In fact, they are so important and distinct that each herd is defined and in most cases named by the use o f a common calving ground. Effective, quantitative methods to measure caribou abundance have only been used since the 1970s and 1980s through aerial surveys. The most recent and effective method of monitoring distribution, which is the key to counting caribou, is through radio- and satellite-telemetry (Hearn et al. 1990; Cameron et al. 1993; Fancy 1994; Gunn et al. 2001). Despite attempts to use longer-term information from archeological sites, explorers and naturalists, company and mission journals, l ive trapping, aerial photos, and radio-telemetry (Elton 1942; Vibe 1967; Krebs et al. 1986; Meldgaard 1986; Messier et al. 1988; Fritz et al. 1993; Couturier et al. 1996), 1 reliable long-term records of caribou population cycles are scarce (Gunn 2005). Recent progress in describing caribou movements and abundance over longer time periods comes from the application of dendroecology. Dendroecology is the study of annual tree-rings as it applies to ecological processes. This method has been an effective tool for reconstructing past population dynamics of voles (Danell et al. 1981), snowshoe hare (Sinclair et al. 1993), porcupines (Spencer 1964; Payette 1987), beavers (Bordage and Fi l ion 1988), and moose (McLaren and Peterson 1994). Source: http://www.nwtwildlife.com/NWTwildlife/caribou/distribution.htm F I G U R E 1.1 The seasonal range and calving grounds of some barren-ground caribou herds in the Northwest Territories and Nunavut. 2 Estimates of vole population sizes were obtained from the amount of bark gnawing on willows (Salix spp.) which roughly reflected vole population density (Danell et al. 1981). Dark bands in a cross-section of a white spruce (Picea glauca (Moench) Voss) were used to date past scars formed by hare browsing (Sinclair et al. 1993). Similarly, the presence o f porcupines was evident from feeding scars on branches and stems of pines (Pinus spp.) (Spencer 1964; Payette 1987). Periods of beaver occupation were determined using a pattern of tree-ring growth release recorded in conifers not selected by beavers during a clearing for the building o f a dam (Bordage and Fi l ion 1988). Finally, ring-width suppression in balsam fir (Abies balsamea var. balsamea), which makes up 59% of the moose's winter diet, was used to indicate increases in moose densities (McLaren and Peterson 1994). Trampling scars left by ungulates (hoofed animals) can be used to reconstruct their population dynamics. Morneau and Payette (1998, 2000) introduced this method to determine the population dynamics of the George River caribou herd in the Quebec-Labrador region. They used trampling scars left by caribou hooves on surficial roots of black spruce (Picea mariana [Mil l . ] BSP) trees along well-used migration trails that were presumed to be quite old. These trampling scars can also be the result of grazing near or on tree roots, where caribou lichens (Cladina rangiferina and other species), a main caribou food source often grow (Morneau and Payette 2000). Scar frequency distributions are then computed based on dendrochronologically-dated scars from root samples. The scar frequency distribution is then interpreted as an index of increasing or decreasing caribou abundance through time. Clearly, using dendroecology to reconstruct caribou abundance differs considerably from previous studies that have adopted dendroecological methods for understanding animal dynamics. This innovative approach as a 3 proxy for caribou abundance is all the more significant given the difficulty and expense of obtaining direct estimates of long-term changes in large mammal population activity. Once the proxy abundance is constructed, the influence of climate on these caribou abundance cycles can be explored. The synchrony in abundance patterns of certain mammalian populations, such as caribou and snowshoe hares, at a regional or continental scale suggests that an external forcing variable is involved (Sinclair et al. 1993; Gunn 2005). Northern biological systems respond to changing climatic regimes over various time scales (Overpeck et al. 1997; Mann and Bradley 1999; Welker et al. 2005). Comparison of fluctuations in caribou abundance at the continental scale suggests a correlation with decadal climate shifts (Gunn 2005). Several studies have linked the patterns o f large-scale climate to mammal abundance. These studies have shown a correlation between abundance patterns in various animal species (snowshoe hares, reindeer, caribou, and muskoxen) and some form o f large-scale climate regime or pattern, such as the Arct ic Oscillation ( A O ; Aanes et al. 2002), the North Atlantic Oscillation ( N A O ; Post et al. 1997; Forchhammer et al. 1998; Post and Stenseth 1999; Post et al. 1999; Forchhammer et al. 2002; Post and Forchhammer 2002) and sunspot cycles (Sinclair et al. 1993). These oscillations have been shown to explain a considerable proportion of annual variations in local temperature and precipitation patterns over large regions (Hurrell 1995), and hence affect mammal abundance through changes in climate. The A O and the N A O are the same physical entity, and are highly correlated however they are considered to be two separate climate oscillations (Thompson and Wallace 1998). The N A O index is calculated as a difference in the normalized monthly mean sea-level pressure between Lisbon, Portugal and Stykkisholmur, 4 Iceland (Hurrell 1996), while the A O index is calculated from the mean monthly sea-level pressure over the northern hemisphere north of 20°N (Thompson and Wallace 2000). The A O explains more than half o f the temperature trends observed i n the eastern Arctic, and less than half o f the temperature trends over the western Arct ic (Rigor et al. 2000). Thus, the A O is more suitable for studying the relation between large-scale climate and mammal abundance in the central N . W . T . The winter index of the A O has shown an upward trend during the last three decades, and is closely related to mean annual temperatures recorded in the Arctic (Overpeck et al. 1997; Thompson and Wallace 1998; Welker et al. 2005; F ig . 1.2). The high index o f the A O is defined as periods o f below normal Arct ic sea-level pressure, as wel l as enhanced surface westerly winds in the north Atlantic. This results in warmer and wetter than normal conditions, and is therefore called the "warm" or positive phase. The low index of the A O is defined as the cooler phase (negative phase), since temperatures tend to be cooler and precipitation tends to be low. Since the A O has been linked to meteorological conditions in the Arctic, it is an ideal region for investigating the interaction between climate variability and ecosystem dynamics (Aanes et al. 2002). Many caribou herds in North America appear to have a population periodicity between 40 and 70 years (Gunn 2005). Similarly, large-scale climate oscillations fluctuate over decades from one mode to another. Thus it is essential that longer timescales are used to study the relation between large-scale climate and caribou. Additionally, sampling across the range o f neighboring herds may provide insight into the synchrony between herds. 5 Sea Level Pressure and Surface Wind AO positive - djf AO negative - djf AZZL Temperature at 2m AO positive AO negative -4 -2 -1 -0.5 -0.25 0 0.25 0.5 1 2 4 C i i i i i i i i -m Source: www.cas.sc.edu/geog/climatelab/climmock.htrnl F I G U R E 1.2 Sea-level pressure and associated temperature anomalies as related to the A O from December-February. The A O is a major mode of circulation variability in the Arctic. 6 In relating large-scale climate to caribou abundance, it is also important to understand the impact of climate at smaller spatial scales. Many studies have quantified the impact of climate on forage quality within caribou ranges (Chapin et al. 1995). Certain climatic conditions have been shown to alter the nutrient concentrations and availability of caribou forage (Chapin and Shaver 1985; Shaver et al. 1986; Chapin et al. 1995), which directly influences body condition o f female caribou (Skogland 1985). Other effects of climate can be seen through changes in the depth and density o f the snow pack which can either reduce or increase rates o f predation (Nelson and Mech 1986; Huggard 1993; Post et al. 1999; Hebblewhite et al. 2002; Hebblewhite 2005). Changes in snow pack characteristics can also affect access to forage for caribou (Turner et al. 1994, Mi l l e r and Gunn 2003). Increased precipitation can result in a greater abundance of mosquitoes and other insects, which has negative effects for caribou as they expend energy trying to avoid mosquitoes by constantly moving as they graze (Downes et al. 1986; Noel et al. 1998). It is the interaction between the Arct ic Oscillation and climate conditions that cause changes in caribou abundance. The following conceptual model illustrates the complex interaction between the Arctic Oscillation, climate and caribou abundance (Fig. 1.3). It is clear that climate has both a direct (e.g. adverse weather) and indirect (e.g. forage quality) effect on caribou abundance. Therefore, understanding how climate affects vegetation is important for providing a better explanation for the relation between climate and caribou. I investigated the impact of regional ( A O index) and local (meteorological data) climate on the variation in tree-ring growth using stems of Picea mariana trees. Aanes et al. 2002 demonstrated that both plant growth (12-year time series) and Svalbard reindeer population growth rate (21-year time series) were negatively related to positive values of the A O index. M y assumption is 7 that i f ring-width growth is limited by certain climate conditions, such as monthly temperature and precipitation, caribou abundance may also be affected either directly or indirectly by the same factors. Arctic Oscillation Positive phase warm/wet winter greater snowdepth, increased predation, decreased access to forage cloudy/cold/wet summer increased abundance of biting insects, sub-optimal growth conditions for forage decreased caribou abundance Negative phase cold/dry winter decreased predation easier access to forage increased caribou abundance F I G U R E 1.3 A conceptual model linking the phases of the Arctic Oscillation, winter and summer climate conditions, and caribou abundance. 1.1 O B J E C T I V E S The main objective of this study was to determine the influence of climate on the abundance cycles of barren-ground caribou (Rangifer tarandus groenlandicus) in the L o w Arctic of central 8 Northwest Territories, Canada. The particular objectives of this study were to: (1) reconstruct the long-term abundance cycles of caribou using frequency distributions of trampling scars on tree roots; (2) evaluate climatic factors correlated with radial growth of Picea mariana trees and develop a climate reconstruction using tree-ring width as a proxy for climate; and (3) assess the impact of large-scale climate regimes (the Arctic Oscillation) on the abundance cycles of these caribou. 1.2 T H E S I S O U T L I N E I compiled this thesis as a series of independent, but related chapters to be submitted for publication in scientific journals. Chapter 1 addresses the main objectives of the research. Chapter 2 describes the reconstruction o f caribou abundance cycles using dendroecological analysis of scar frequencies from roots of Picea mariana and has been accepted for publication in Arctic, Antarctic and Alpine Research (In Press, September 2006). In Chapter 3, I present a series of seven tree-ring chronologies developed from the same sites as those used to reconstruct the caribou abundance cycles from Chapter 2.1 also determined the main climatic factors driving tree-ring growth and reconstructed July-August temperatures using five of the longest chronologies. Chapter 4 was dedicated to determining the relation between the Arctic Oscillation and caribou abundance from 1900-2002. The final chapter is a summary and synthesis of the major findings of this research. 9 1.3 R E F E R E N C E S Aanes, R., Saether, B . , Smith, F . M , Cooper, E.J . , Wookey, P .A . , and 0ritsland, N . A . 2002. The arctic oscillation predicts effects of climate change in two trophic levels in a high-arctic ecosystem. Ecology Letters, 5: 445-453. Bordage, G . , and Fi l ion, L . 1988. Analyse dendroecologique d'un milieu riverain frequente par le castor (Castor canadensis) au Mont D u Lac-Des-Cygnes (Charlevoix, Quebec). Naturaliste canadien, 115: 117-124. Cameron, R .D . , Smith, W.T. , Fancy, S.G., Gerhart, K . L . , and White, R . G . 1993. Calving success of female caribou in relation to body weight. Canadian Journal of Zoology, 71: 480-486. Chapin, F.S., III, Shaver, G.R., Gibl in , A . E . , Nadelhoffer, K . , and Laundre, J .A. 1995. Responses of arctic tundra to experimental and observed changes in climate. Ecology, 76: 694—711. Chapin, F.S., III, and Shaver, G.R. 1985. Individualistic growth response of tundra plant species to environmental manipulations in the field. Ecology, 66: 564-576. Couturier, S., Courtois, R., Crepeau, FL, Rivest, L . -P . , and Luttich, S.N. 1996. The June 1993 photocensus of the Riviere George caribou herd and comparison with an independent census. Rangifer, Special Issue no., 9: 283-296. Danell, K . , Ericson, L . , and Jakobsson, K . 1981. A method for describing former fluctuations of voles. Journal of Wildlife Management, 45(4): 1018-1021. Downes, C M . , Theberge, J .B. , and Smith, S . M . 1986. The influence o f insects on the distribution, microhabitat choice, and behavior of the Burwash caribou herd. Canadian Journal of Zoology, 64: 622-629. 10 Elton, C . 1942. Voles, mice and lemmings: problems in population dynamics. London: Oxford University Press, pp. 496. Fancy, S.G., Whitten, K . R . , and Russell, D . E . 1994. Demography o f the Porcupine caribou herd, 1983-1992. Canadian Journal of Zoology, 72: 840-846. Forchhammer, M . C . , Post, E . , Stenseth, N . C . , and Boertmann, D . M . 2002. Long-term responses in arctic ungulate dynamics to changes in climatic and trophic processes. Population Ecology, 44: 113-120. Forchhammer, M . C . , Stenseth, N . C . , Post, E. , and Langvatn, R. 1998. Population dynamics o f Norwegian red deer: density-dependence and climatic variation. Proceedings of the Royal Society of London Bulletin, 265: 341-350. Fritz, R., Suffling, R., and Younger, T . A . 1993. Influence o f fur trade, famine, and forest fires on moose and woodland caribou population in northwestern Ontario from 1786 to 1911. Environmental Management, 17:477-489. Gunn, A . 2005. Voles, lemmings and caribou - population cycles revisited? Rangifer Special Issue, 14: 105-112. Gunn, A . , Dragon, J. and Boulanger, J. 2001. Seasonal movements o f satellite-collared caribou from the Bathurst herd. Final Report to the West Kitikmeot Slave Study Society, Yellowknife, N W T . http://www.wkss.nt.ca/HTML/08_ProiectsReports/PDF/SeasonalMovementsFinal.pdf Gunn, A . , and Fournier, B . 2000. Identification of caribou calving grounds on the NWT mainland and islands. Yellowknife, N W T : Department of Resources, Wildlife and Economic Development, Government of Northwest Territories, Fi le Report #123. 11 Hearn, B . J . , Luttich, S .N. , Crete, M . , and Berger, M . B . 1990. Survival of radio-collared caribou (Rangifer tarandus caribou) from the George River herd, Nouveau-Quebec-Labrador. Canadian Journal of Zoology, 68: 276-283. Hebblewhite, M . 2005. Predation by wolves interacts with the North Pacific Oscillation (NPO) on a western North American elk population. Journal of Animal Ecology, 74: 226-233. Hebblewhite, M . , Pletscher, D . H . , and Paquet, P .C. 2002. E lk population dynamics in areas with and without predation by recolonizing wolves in Banff National Park, Alberta. Canadian Journal of Zoology, 80: 789-799. Huggard, D . J . 1993. Effect of snow depth on predation and scavenging by gray wolves. Journal of Wildlife Management, 57: 382-388. Hurrell , J . W . 1995. Influence o f variations in extratropical wintertime teleconnections on Northern Hemisphere temperature. Geophysical Research Letters, 23: 665-668. Hurrell , J . W . 1995. Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation. Science, 269: 676-679. Krebs, C . J . , Gilbert, B .S . , Boutin, S., Sinclair, A . R . E . , and Smith, J . N . M . 1986. Population biology o f snowshoe hares. I. Demography of food-supplemented populations in the southern Yukon, 1976-84. Journal of Animal Ecology, 55(3): 963-982. Mallory, F .F. , and Hi l l i s , T .L . 1998. Demographic characteristics of eircumpolar caribou populations: ecotypes, ecological constraints, releases, and population dynamics. Rangifer, Special Issue #10, 49-60. Mann, M . E . , and Bradley, R.S . 1999. Northern Hemisphere temperatures during the past millennium: inferences, uncertainties, and limitations. Geophysical Research Letters, 26(6): 759-762. 12 McLaren, B . E . , and Peterson, R .O . 1994. Wolves, moose, and tree rings on Isle Royale. Science, 266:1555-1558. Meldgaard, M . 1986. The Greenland caribou: zoogeography, taxonomy, and population dynamics. Meddelelser om Grenland, BioScience, 20: 1-88. Messier, F., Huot, J., Le Henaff, D. , and Luttich, S. 1988. Demography of the George River caribou herd: evidence of population regulation by forage exploitation and range expansion. Arctic, 41 (4): 279-287. Mi l le r , F . L . , and Gunn, A . 2003. Catastrophic die-off o f Peary caribou on the Western Queen Elizabeth Islands, Canadian High Arctic. Arctic, 56: 381-390. Morneau, C , and Payette, S. 2000. Long-term fluctuations of a caribou population revealed by tree-ring data. Canadian Journal of Zoology, 78: 1784-1790. Morneau, C , and Payette, S. 1998. A dendroecological method to evaluate past caribou (Rangifer tarandus L.) activity. Ecoscience, 5(1): 64-76. Nelson, M . E . , and Mech, L . D . 1986. Relationship between snow depth and gray w o l f predation on white-tailed deer. Journal of Wildlife Management, 50: 471-474. Noel , L . E . , Pollard, R . H . , Ballard, W . B . , and Cronin, M . A . 1998. Act ivi ty and use o f active gravel pads and tundra by caribou, Rangifer tarandus granti, within the Prudhoe Bay O i l Field, Alaska. Canadian Field-Naturalist, 112: 400-409. Overpeck, J., Hughen, K . , Hardy, D . , Bradley, R., Case, R., Douglas, M . , Finney, B . , Gajewski, K . , Jacoby, G . , Jennings, A . , Lamoureux, S., Lasca, A . , MacDonald, G . , Moore, J., Retelle, M . , Smith, S., Wolfe, A . , and Zielinski, G . 1997. Arctic environmental change of • the last four centuries. Science, 278: 1251-1256. 13 Payette, S. 1987. Recent porcupine expansion at tree line: a dendroecological analysis. Canadian Journal of Zoology, 65: 551-557. Post, E . and Forchhammer, M . C . 2002. Synchronization of animal population dynamics by large-scale climate. Nature, 42: 168-171. Post, E . , Peterson, R .O. , Stenseth, N . C . , and McLaren, B . E . 1999. Ecosystem consequences of wol f behavioural response to climate. Nature, 401: 905-907. Post, E . , Stenseth, N . C . , Langvatn, R., and Fromentin, J . - M . 1997. Global climate change and phenotypic variation among red deer cohorts. Proceedings of the Royal Society of London Bulletin, 264: 1317-1324. Post, E . , and Stenseth, N . C . 1999. Climatic variability, plant phenology, and northern ungulates. Ecology, 80: 1322-1339. Rigor, I.G., Colony, R . L . and Martin, S. 2000. Variations in surface air temperature observations in the Arctic, 1979-97. Journal of Climate, 13: 896-914. Shaver, G.R. , Chapin, F.S., III, and Gartner, B . 1986. Factors limiting seasonal growth and peak biomass accumulation in Eriophorum vaginatum in Alaskan tussock tundra. Journal of Ecology, 74: 257-278. Skogland, T. 1985. The effects of density-dependent resource limitation on the demography of wi ld reindeer. Journal of Animal Ecology, 54: 359-374. Sinclair, A . R . E . , Gosline, J . M . , Holdsworth, G . , Krebs, C.J . , Boutin, S., Smith, J.N.R., Boonstra, R., and Dale, M . 1993. Can the solar cycle and climate synchronize the snowshoe hare cycle in Canada? Evidence from tree rings and ice cores. American Naturaliste, 141(2): 173-198. 14 Spencer, D . A . 1964. Porcupine population fluctuations in past centuries revealed by dendrochronology. Journal ofApplied Ecology, 1(1): 127-149. Thompson, D .W.J . , and Wallace, J . M . 2000. Annular modes in the extratropical circulation, part I: Month-to-month variability. Journal of Climate, 13: 1000-1016. Thompson, D.W.J . , and Wallace, J . M . 1998. The Arct ic Oscillation signature in the wintertime geopotential height and temperature fields. Geophysical Research Letters, 25(9): 1297-1300. Turner, M . G . , W u , Y . , Wallace, L . L . , Romme, W . H . , and Brenkert, A . 1994. Simulating winter interactions among ungulates, vegetation, and fire in northern Yellowstone Park. Ecological Applications, 4: 472-496. Welker, J . M . , Rayback, S., and Henry, G.H.R. 2005. Arctic and North Atlantic Oscillation phase changes are recorded in the isotopes ( d l 8 0 and d l 3 C ) o f Cassiope tetragona plants. Global Change Biology, 11: 997-1002. Vibe, C. 1967. Arctic animals in relation to climatic fluctuations. Meddelelser om Grenland, 170: 1-227. 15 CHAPTER 21. LONG-TERM ABUNDANCE PATTERNS OF BARREN-GROUND CARIBOU USING TRAMPLING SCARS ON ROOTS OF PICEA MARIANA IN THE NORTHWEST TERRITORIES 2.1 INTRODUCTION Barren-ground caribou {Rangifer tarandus groenlandicus) tend to undergo relatively regular changes in population size over decadal time scales (Skoog 1968). In the Northwest Territories, there are seven distinct barren-ground caribou herds. During August and September, the caribou move south from the tundra and into the forest-tundra transition zone along the treeline (Gunn et al. 2001). During the spring, caribou migrate to their calving grounds, which can be as far as 700 km away from their wintering grounds. Barren-ground caribou adult males stand about 115 cm high and weigh approximately 105 kg (Kelsall 1968). The average life span o f the barren-ground caribou is between 10-15 years. The population o f the barren-ground caribou herds in the studied region range between 186,000 (2003 estimate of the Bathurst herd) to 276,000 (1994 estimate of the Beverly herd) (http://www.nwtwildlife.com/NWTwildlife/caribou/herds.htm). Aerial surveys have provided accurate data on caribou numbers over the past few decades (e.g. Gunn 2005); however, longer-term information is less available. Aboriginal knowledge o f caribou populations does extend further back in time and has been compiled for the Bathurst herd in the Northwest Territories, Canada (Dogrib Treaty 11 Council 2001). For other herds such as the Beverly herd, whose range is adjacent to the Bathurst herd, the traditional aboriginal information is not well compiled and census data are fewer. ' A version of this chapter has been accepted for publication in Arctic, Antarctic and Alpine Research (In Press, September 2006). Zalatan, R., Gunn, A. and Henry, G.H.R. 2006. Long-term abundance patterns of barren-ground caribou using trampling scars on roots of Picea mariana in the Northwest Territories, Canada. 16 Caribou abundance among different herds has been shown to be in synchrony at continental scales (Gunn 2005). This synchrony among herds in Alaska, Greenland, and eastern North America has been well studied (Gunn 2005). However, knowledge of long-term abundance cycles for the central Canadian barren-ground caribou herds is not well understood. The only data that researchers and managers have to work with is derived from aerial photography and traditional knowledge (TK) . Caribou are large-bodied animals whose longevity necessitates a dataset that covers a larger geographic area and a longer timescale (Gunn 2005). A recent method to estimate changes in caribou abundance over decades is based on the application of dendroecology. Morneau and Payette (1998, 2000) and Boudreau et al. (2003) used dendroecology to describe changes in the size of the George River caribou herd in the northern Quebec-Labrador region. The authors aged the scars left by caribou hooves on the top of surficial roots or low branches of spruce trees during their summer migration to the tundra. The scars are formed when part of the bark is removed due to trampling, which causes cambium death and stops radial growth in that section of the root. A scar lobe forms around the damaged cambial tissue in subsequent years. The date of scar formation can be accurately determined using dendroecology, and frequency distributions are then computed from the dated scars on root and stem samples. This method has made it possible to reconstruct caribou population activity and provides the longest available proxy record for changes in caribou herd size. The objective of this study was to evaluate the long-term population dynamics of barren-ground caribou herds in the central Northwest Territories, Canada through the use of dendroecology on trampling scars from spruce stands in the forest-tundra. Three aspects of the methodology were 17 considered to properly evaluate the use of these data for reconstruction of caribou population dynamics: (1) the effect of root age on the scar frequency distribution; (2) the synchronicity of populations in neighbouring herds; and, (3) a comparison of the scar frequency distribution to historical and recent data (from aerial photography) available for the Bathurst herd. 2.2 M E T H O D S 2.2.1 Study Sites The climate of this region is characterized by long, cold winters and short, cool summers with mean January and July temperatures of -26.8°C and 16.8°C, respectively (Meteorological Service of Canada 1971-2000 climate normals from Yellowknife airport, 62°27' N , 114°26' W, 206 m a.s.l.). Precipitation is low with an average annual total of 280.7 mm, falling mostly as rain during the summer months. The landscape consists of broad uplands and shallow lowlands, with rock outcrops, hummocky and ridged morainal deposits and eskers, and numerous lakes and wetlands (Traynor 2001). Permafrost is generally continuous. The vegetation consists of open lichen woodland with patches of closed forest in the southern areas, to forest-tundra consisting of thinning patches of forest and shrub tundra, to low arctic tundra in the north (Matthews et al. 2001). The 19 study sites were located along the treeline to the northwest and southeast of Yellowknife (between 65°09'N, 115°37'W and 61°32'N, 105°52'W; Fig. 2.1). The sampling took place in the forest-tundra (Fig. 2.2). Fifteen study sites were selected based on information from Dogrib elders who identified several regions across the treeline within the Bathurst herd's annual range 18 that were frequented by caribou (Dogrib Treaty 11 Council 2001; Fig. 2.1, and Fig. 2.3). Figure 2.4 depicts a group of Bathurst caribou migrating across the landscape in the N.W.T. 125°W 120°W 115°W Great Bear L. 100 I •12 / 11 •10 14-15 * L2 [Northwest sites / Yellowknife Elevation (m) 0-100 101-300 301 - 500 501 - 700 701 - 950 • Study sites — Treeline — • Migratory route! 19 J 6 , 200 km —J Southeast sites 115°W 110°W 17 105°W h65°N F I G U R E 2.1 Sampling sites of scarred roots on Picea mariana (sampled 2002) and the summer migratory route of barren-ground caribou. The geographic coordinates of the sampling locations for the scarred roots are listed in Table 2.1. A further four sites were selected on the late summer range of the Beverly herd. A targeted sampling approach was essential to find black spruce (Picea mariana [Mill.] BSP) roots with 19 trampling scars caused by caribou. To minimize bias in this study, sampling was done a few meters off the caribou trails to establish the presence of trampling scars on roots growing across trails not currently used by caribou. One site was selected for this purpose. Laboratory analysis revealed that trampling scars were not found in the samples collected off the caribou trails. A s a result, this method of sampling off trails was not applied to any subsequent sites. F I G U R E 2.2 This is an example of a typical site where caribou scars were collected. This photo was taken from a helicopter (at approximately 300 m elevation) and is located in the forest-tundra within the annual range of the Bathurst caribou herd. Once the general location o f the sites was chosen, a series of highly trampled caribou trails were randomly chosen per site for sampling (Fig. 2.5). External features such as exposed xylem and resin accumulation were used to identify roots with scars. Roots found to be buried under understorey plants were all checked for possible scarring. Based on the density of trails, the 20 number of roots sampled per site differed. The roots sampled were all at the ground surface and therefore digging was not necessary. Source: R. Zalatan FIGURE 2.3 Caribou migration trails across an esker taken from a helicopter in the central N . W . T . Source: R. Zalatan FIGURE 2.4 Bathurst caribou migrating across the landscape in the N . W . T . 21 Source: R. Zalatan F I G U R E 2.5 A series of black spruce roots trampled by barren-ground caribou. Trampling removes part of the bark, leaving behind a scar which is used to represent caribou abundance. Selecting sites used by barren-ground caribou populations increased our chances of finding old scars, thereby lengthening the scar chronology. The sites in this study cover a portion of the summer range of the Beverly, Bathurst and Ahiak herds, therefore the caribou abundance cycles derived from the scar frequency distribution are not defined as those of a specific herd. The analysis was performed on two groups of sites: (1) the northwest sites (n = 15); and, (2) the southeast sites (n = 4). These groups correspond generally to the range of the Bathurst and Beverly caribou herds, respectively. However, there can be considerable overlap in the ranges of 22 these herds (http://www.nwtwildlife.rwed.gov.ntxa/images/new_pa6.gi f) in some years; therefore, the sites w i l l be referred to as the northwest and southeast sites. Tracks of satellite-collared caribou cows from the Bathurst herd over the period from 1996-2002 clearly demonstrate that the area where the study sites are located has been well used by caribou (Fig. 2.6). Source: A. Gurm 2002 F I G U R E 2.6 Tracks o f 10-20 satellite-collared caribou cows from the Bathurst herd (1996-2002). The black boxes outline the general area where the study sites are located. 23 A t each study site, trampled caribou trails were selected and surficial roots of Picea mariana trees were collected where they crossed heavily trampled caribou paths. The number of roots sampled per site varied from 10 to 417, depending on the density of trails and the time available for sampling. For most o f the sampling, only one root was cut per tree. Caribou scars are formed on the top o f surficial roots and sometimes on low branches of krummholz trees during the snow-free period. The shape of the scar is generally round, elliptical, or elongated with neat margins (Fig. 2.7). scar year o f scar formation pith Source: R. Zalatan F I G U R E 2 . 7 Schematic diagram of a scar on a cross-sectional sample o f a root. Scars can be formed by other ungulates or human activity (such as hiking trails) or by fire, however there was little evidence that the study region has been affected by any o f these disturbance agents other than caribou. The scars found in this study were clearly trampling scars 24 because charcoal was not detected and the samples were taken on roots and not on the stems of trees. In addition, fire (Barrett and Arno 1988) and frost scars are structured differently and are therefore easy to differentiate from trampling scars. Cross-sections were sanded using progressively finer grades of sandpaper (240, 320, 400, and 600) and were prepared for crossdating as outlined in Stokes and Smiley (1968). Each annual ring was measured using a Mini-scale and Mate System Encoder connected to an AcuRite III digital counter with a precision of 0.001 mm. Scars were dated by visually and statistically crossdating the rings prior to the scar. Crossdating ensures the exact year of formation of annual rings and it verifies the presence of missing, false or locally absent rings (Fritts 1976). Most scars were dormant season scars, which were formed in late summer/early fall when caribou migrate through the study sites upon returning from the calving grounds. However, it is possible during a winter with low snow-pack, that some scars are formed in the spring when caribou are on their way to the calving grounds. The seasonal dormant phase o f the cambium extends over 2 calendar years (Morneau and Payette 2000). It is not possible to determine the exact date that scars formed, as the year of scar formation could vary by +1 year. A s a standard, the year o f scar formation is taken as the most recent year (Morneau and Payette 1998). A t times, annual rings were wedged (or pinched) near the edge o f the scar, which made it difficult to identify the exact scar ring. A s a consequence, we grouped the scars into 5-year age-classes to obtain the most accurate record. Only scars that were successfully crossdated were included in subsequent analysis. 25 Once the date of scar formation was determined, the scar frequency distribution (5-year age-classes) was computed as an index of caribou abundance (Morneau and Payette 1998, 2000). Although the scars are accurately dated, they are representing caribou abundance and therefore a 5-year age-class is a more conservative way of capturing any variability in this index of caribou abundance. The scar frequency distribution must be interpreted in terms of successive periods of increasing and decreasing caribou abundance. Scar frequency distributions were also computed using scars at an annual level, a 2-year age-class, a 3-year age-class and a 4-year age-class to determine whether there were any significant changes in the trends of the scar frequency distributions for both groups of sites (northwest and southeast sites; Appendix A ; Fig . A 1 - A 4 ) . Spearman's rank correlations were computed between the two groups of sites for each age-class (Appendix A ; Table A . l ) . A Kolmogorov-Smirnov test for goodness-of-fit was used to determine i f the scar frequency distributions for both herds were significantly different. To ensure the validity of the results, the scar frequency was compared to traditional knowledge (TK) and animal counts from aerial surveys and photography. Research on the T K data was conducted by the West Kitikmeot Slave Study in conjunction with the aboriginal elders of the Dogrib Nation. The data were used to index Bathurst caribou herd abundance to the 1920s (Dogrib Treaty 11 Council 2001). The T K data were only recorded back to the 1920s because the elders would not provide any information on caribou population abundance and migration that they had not directly witnessed or for which they felt certain about. 26 Aerial surveys of caribou herds in the Northwest Territories are performed by the Department o f Environment and Natural Resources (ENR) . Biologists estimate the number of breeding females in a herd by performing aerials surveys above the calving grounds. Since all breeding females calve around the same time in early June and in the same general area, it makes it possible to ensure that all the animals counted belong to a single herd. The percentage of the total herd that is breeding females is used to estimate the size of the entire herd. A series of photos (e.g. 2600 photos were taken for the 2003 estimate) are taken while flying at 600 meters above ground level (http://www.nwtwildlife.com/NWTWildlife/caribou/Counting_Caribou.pdf). A standard error is calculated for each year that a survey is done. For the reconstruction o f caribou abundance using dendroecology, two analyses were performed. The first involved determining the influence of root age to obtain an objective scar frequency distribution independent of the age of the root itself. This was done by plotting the scar frequency distribution using only roots that had established prior to 1900 (the year to which the data set was eventually truncated). This analysis was used to account for the increasing underestimation of caribou abundance with the increasing age o f the roots. A second analysis was used to address the underestimation of caribou activity with time, assuming a constant loss of scars. Various factors contribute to the loss of scars through time such as the death o f scar-bearing roots, fading of scars by weathering, decomposers, and repeated caribou trampling activity (Morneau and Payette 2000). A log-linear regression was applied to remove the error associated with the loss of scars through time and the residuals were used to obtain a more accurate depiction of the abundance patterns of these herds. Departures from the negative exponential model were used to demonstrate the years of high and low caribou abundance. 27 2.3 R E S U L T S 2.3.1 Scar frequency distributions Statistical and visual crossdating of 2477 root samples collected from Picea mariana trees yielded 1991 trampling scars (Table 2.1). T A B L E 2.1 Geographic location, number of scars and samples for the northwest and southeast sites, Northwest Territories. Locations are also shown in F ig . 2.1. Location Site Number of Scars Number of Scars (% of all sites) Number of Samples Number of Samples (% of all sites) Latitude Longitude Northwest sites 1 61 3.1 36 1.5 64° 27.2' N 112° 45.6'W 2 19 1.0 21 1.1 64° 27.9' N 113° 09.4'W 3 12 0.6 10 0.5 64° 22.7' N 113° 23.3'W 4 8 0.4 12 0.6 64° 17.5'N 113° 49.3'W 5 62 3.1 108 5.4., 63° 29.1'N 112° 23.2'W 6 190 9.5 282 14.2 63° 32.7' N 112° 18.8'W 7 47 2.4 143 7.2 63° 21.9' N 111° 46.2' W 8 42 2.1 139 7.0 63° 13.9'N 110° 55.2'W 9 41 2.1 100 5.0 64° 04.8' N 112° 31.1* W 10 34 1.7 72 3.6 64° 37.7' N 115° 04.4'W 11 90 4.5 136 6!8 64° 47.7' N 115° 12.2'W 12 125 6.3 121 6.1 65° 00.8'N 115° 29.1'W 13 77 3.9 93 4.7 65° 09.8'N 115° 37.5'W 14 37 1.9 32 1.6 64° 30.8'N 114° 15.0'W 15 102 5.1 153 7.7 64° 30.5'N 114° 14.6'W Total 947 1458 Southeast sites 16 225 11.3 245 12.3 61°45.1'N 106° 29.0'W 17 321 16.1 282 14.2 61° 32.5' N 105° 52.3'W 18 436 21.9 417 20.9 61° 36.5' N 106° 07.3'W 19 62 3.1 75 3.8 61° 36.0'N 106° 49.3'W Total scars 1044 1019 Total (all sites) 1991 100 2477 100 28 The number of scars found was evenly distributed between the two groups o f sites: 48% (n = 947) within the northwest sites (sites 1-15) and 52% (n = 1044) within the southeast sites (sites 16-19). For all sites, 97% (n =1933) of the scars were dated with certainty. The earliest scars in our samples were formed in the northwest sites during the 1760s, although they were scarce until the 1870s (Fig. 2.8a). The scars from the southeast sites appear in the 1835 age-class. The majority (97%) o f the scars were formed after 1900, although, no scars formed from 1910-1930 in the northwest sites. The frequency of scars increased steadily through time with the highest frequency o f scars in the 1945, 1990 and 1995 age-classes, in addition to the 1975 age-class for the southeast sites. There was a general trend toward increasing scars through time until the present, with a sudden drop in the number o f scars in 2000. Similar scar frequency distributions were observed at both groups of sites (Spearman's rank correlation, r s = 0.88; p<0.05, n = 24), and the scar frequency distributions were not different ( K S , p>0.05). The cumulative frequency curves of the number o f samples in Figure 2.8a were used to demonstrate how the number of samples increased steadily through time. The cumulative frequency curves were not applied to Figure 2.8b because this figure did not include the entire dataset. To assess the influence of age o f roots on the scar data, the scar frequency distribution was plotted using only roots established prior to 1900 present over the entire period of the truncated scar frequency distribution (Fig. 2.8b). Thus, the relative frequency o f scars is compared with the number of roots that existed from 1900 to the present. The scar frequencies for both groups of 29 (a) 15 10 m x Northwest sites Southeast sites 8 0 60 S 40 o 20 1760 1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 Year (5-year age-classes) (b) 12 10 a. 4 1 Northwest sites Southeast sites Lliifi/jlii 0 1760 1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000 Year (5-year age-classes) F I G U R E 2.8 Scar frequency distribution (a) using all samples from sites 1-15 (blue bars) and sites 16-19 (grey bars) and cumulative frequencies of the number of samples (lines) (r s = 0.88; p<0.05, n = 24); and (b) using only scars on roots established before 1900 (r s = 0.63, p<0.05, n = 24). 30 sites, showed the same patterns of major increases and decreases (Spearman's rank correlation, (r s = 0.63; /?<0.05, n = 24). Both groups o f sites showed low numbers until the 1920s, again from 1955-1970 and in 2000. The scar frequency distributions showed an increase during the 1940s until the mid-1950s, and from 1980-2000. However, the complete scar frequency distribution (Fig. 2.8a) showed a much more marked peak in the 1990s than the restricted distribution (Fig. 2.8b). The scar frequency distribution from the northwest sites (corresponding to the Bathurst herd; truncated to 1900) was compared to information on caribou abundance from Dogrib elders and from animal counts based on aerial photography (Fig. 2.9). A similar calibration of the data for the southeast sites (Beverly herd) was not possible because aboriginal knowledge has not been compiled and there have been fewer population estimates based on aerial photographic surveys. The qualitative description of "high" or " low" population of caribou was derived from aboriginal elder's narratives and describes the abundance of caribou from the 1920s to the 1970s (Dogrib Treaty 11 Council 2001). The population estimates were from aerial photographic surveys and describe caribou abundance from 1984 until 2003. The information from the Dogrib elders and the scars both showed low numbers of caribou during the 1920s, followed by a high peak in caribou numbers during the mid-1940s, then a low period from 1950 to 1970. The aerial photography data showed an increasing trend in caribou abundance after the 1970s, with a peak in the mid-1980s, followed by a significant drop in caribou abundance in 2000. In the 1990s, the scar frequency distribution 31 showed a rapid increase, after a period of little change from the 1950s-70s, followed by a notable decrease after 2000. 500000 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Year (5-year age-classes) F I G U R E 2.9 Bathurst herd caribou numbers from aerial photography surveys (1984-2003; black line; means ± SE), and abundance patterns (high = dark grey shading or low = light grey shading) derived from traditional knowledge of elders from the Dogrib First Nation relative to the scar frequency distribution for the northwest sites (Bathurst herd; bars). 2.3.2 Ca r ibou abundance cycles The scar frequency distributions for both groups of sites demonstrated an exponential decrease in scars from the 2000 age-class towards the 1760s. Figure 2.10 illustrates the residuals of the log-linear regression for both groups of sites, which was used to remove the long-term depletion pattern in the number of scars and to accentuate the fluctuations in the scar frequency distributions. 32 1900 1920 1940 1960 1980 Year of scar formation (5-year age-classes) 2000 F I G U R E 2.10 Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups o f sites. The large oscillations in the residuals around the regression line follow a pattern similar to that of the fluctuations in the scar frequency distributions. The chronologies o f residuals from both sites were positively correlated (Spearman's rank, r s = 0.55; p<0.05). The correlations remained significant when the recent (1950-2005) and older (1900-1950) parts of the chronologies were considered separately (Spearman's rank, r s = 0.55 and r s = 0.87; p<0.05, respectively). 2.4 D I S C U S S I O N The scar frequency distribution developed in this study is the longest proxy record of barren-ground caribou abundance yet available. The proxy record developed in this study extends caribou abundance data back to the 1760s, although the scars were scarce until approximately 33 1900. The most prominent peaks in the scar frequency distribution were seen during the mid-19408, and the late 1980s and 1990s. Periods o f low scar frequency occurred in the 1920s, 1950s-70s and at the turn of the 21 s t century. The loss of scars with time did not influence the population reconstruction since the scar frequency distribution using only roots established before 1900 demonstrated a broadly similar abundance pattern. Both groups o f sites showed similar patterns in scar frequency, indicating that the different herds within the sampling area experienced similar changes in abundance over time. Given the scale of these synchronous changes in caribou abundance, it is l ikely they are linked to changes and variability in large-scale climate, such as the Arct ic Oscillation (Aanes et al. 2002; Post and Forchhammer 2002; Post and Forchhammer 2004). The link between the abundance patterns and climate is investigated in Chapter 4. High numbers o f scars were associated with the growth of the herds between the mid-1940s to the 1990s, as seen from the Dogrib traditional knowledge (TK) and aerial photography data. The trends seen in data from T K are similar to those depicted in the scar frequency distribution from 1920-1970. High numbers of caribou during the 1980-1990s and sudden drop in the 2000 age-class were observed in the aerial photography data. The large standard error associated with the aerial photography data may account for the earlier peak observed during the mid-1980s (SE±72,000 in 1986). This provides evidence for the strength and accuracy of the scar frequency distribution as a proxy for caribou abundance. Despite the earlier peak in the mid-1980s from the census data, there were synchronous patterns in caribou abundance using the different methods. Morneau and Payette (2000) found a similar synchronous relationship between estimates of the herd size and trampling scar data for the George River caribou herd, with increasing caribou 34 numbers from the early 1940s until the early 1990s. Synchronous changes in the scar frequency distribution at both groups of sites and among the different datasets demonstrate the strength of the spatiotemporal pattern in caribou abundance in this region. Changes in caribou migration patterns have not been quantified over the last 100 years, however, the synchrony in the scar frequency data to the T K and aerial photography data, suggests that any changes in migratory routes has not affected the results, Wi th the use of dendroecology, I have shown that it is possible to reconstruct the abundance patterns o f these barren-ground caribou herds. The presence of trampling scars at all sites suggests that the forest-tundra is an ideal region for estimating past caribou activity. Assuming the capacity o f conifer roots to produce scars remains constant with time (a reasonable assumption), the scar frequencies can be used as a proxy for caribou abundance (Morneau and Payette 1998, 2000). A log-linear regression was applied to account for the loss of scars with time, assuming that this loss is constant (Morneau and Payette 2000). The residuals o f the regression were then used to depict the abundance patterns of the barren-ground caribou at these sites. The chronology of residuals illustrated the overall trends in caribou activity as seen in the scar frequency distributions. The similar trends in the T K , caribou numbers and trampling scars suggest that changes in the scar frequency distributions corresponded to changes in the rate of scar formation. Subsequently, variations in the rate of scar loss with time, which corresponds to caribou movements along trails and root mortality, were most l ikely minor in comparison to the changing rate of scar formation. 35 This study further validates the use of dendroecology for understanding the population dynamics o f barren-ground caribou herds (Morneau and Payette 1998, 2000). The identification of similarities in the scar frequency distributions among sites can lead to further investigation regarding the spatial and temporal patterns throughout the range o f the barren-ground caribou herds. The sites selected for this study represent a substantial portion of the range of barren-ground caribou. However, the length and lack of synchrony in the scar frequency prior to 1900 needs to be addressed with further sampling and analysis. Nevertheless, further use of this method has the potential of providing valuable information about the long-term abundance cycles of all barren-ground caribou herds across northern North America. This study was the first to reconstruct long-term abundance cycles for barren-ground caribou of the Northwest Territories. 36 2.5 R E F E R E N C E S Aanes, R., Saether, B - E . , Smith, F . M . , Cooper, E.J . , Wookey, P .A . , and 0ristland, N . A . 2002. The Arctic Oscillation predicts effects o f climate change in two trophic levels in a high-arctic ecosystem. Ecology Letters, 5: 445-453. Barrett, S.W., and Arno, S.F. 1988. Increment-borer methods for determining fire history in coniferous forests. General Technical Report, Report INT-244. Ogden, U T : Intermountain Research Station. Boudreau, S., Payette, S., Morneau, C , and Couturier, S. 2003. Recent decline of the George River caribou herd as revealed by tree-ring analysis. Arctic, Antarctic, and Alpine Research, 35: 187-195. Dogrib Treaty 11 Council . 2001. Caribou migration and the state of their habitat. Final Report to the West Kitikmeot/Slave Study Society, Yellowknife, N T , 90 pp. Fritts, H . C . 1976: Tree Rings and Climate. New York: Academic Press, 567 pp. Gunn, A . , Dragon, J., and Boulanger, J. 2001. Seasonal movements of satellite-collared caribou from the Bathurst herd. Final Report to the West Kitikmeot Slave Study Society, Yellowknife, N W T , 72 pp. http://www.wkss.nt.ca/HTML/08_ProjectsReports/PDF/SeasonalMovementsFinal.pdf Gunn, A . 2005. Voles, lemmings and caribou - population cycles revisited? Rangifer Special Issue, No . 14: 105-112. Kelsal l , J.P. 1968. The migratory barren-ground caribou of Canada. Ottawa: Queen's Printer, 340 pp. 37 Matthews, S., Epp, FL, and Smith, G . 2001. Vegetation classification for the West Kitikmeot/Slave Study region. Final report to the West Kitikmeot/Slave Study. Society, Yellowknife, N T , 42 pp. Morneau, C. , and Payette, S. 1998. A dendroecological method to evaluate past caribou (Rangifer tarandus L.) activity. Ecoscience, 5: 64-76. Morneau, C. , and Payette, S. 2000. Long-term fluctuations of a caribou population revealed by tree-ring data. Canandian Journal of Zoology, 78: 1784-1790. Post, E . , and Forchhammer, M . C . 2002. Synchronization of animal population dynamics by large-scale climate. Nature, 420: 168-171. Post, E . , and Forchhammer, M . C . 2004. Spatial synchrony of local populations has increased in association with the recent Northern Hemisphere climate trend. Proceedings of the National Academy of Sciences of the United States of America, 101: 9286-9290. Skoog, R .O . 1968. Ecology of the caribou (Rangifer tarandus grand) in Alaska. Ph.D. Thesis, University of California, Berkeley. Stokes, M . A . , and Smiley, T . L . 1968. An Introduction to tree ring dating. Chicago: University of Chicago Press, 73 pp. Traynor, S. 2001. Esker habitat characteristics and traditional use study in the Slave Geological Province. Final report to the West Kitikmeot/Slave Study Society. Yellowknife: West Kitikmeot/Slave Study Society, 66 pp. 38 CHAPTER 3. SPATIO-TEMPORAL VARIATION OF TREE-RING GROWTH ALONG A TRANSECT AT TREELINE WITHIN THE RANGE OF BARREN-GROUND CARIBOU: A DENDROCLIMATIC APPROACH 3.1 INTRODUCTION The region to the northwest and southeast of Great Slave Lake is home to four barren-ground caribou herds, whose movements are monitored through satellite telemetry as they migrate from the boreal forest across the latitudinal treeline into the tundra (http://www.nwtwildlife.com/). Studies have demonstrated the direct impacts of climate at different scales on the migratory patterns and dynamics o f this ungulate species. For example, Telfer and Kelsal l (1984) found that an abundance of soft snow made it easier for caribou to avoid predation. A case-study on the Peary caribou demonstrated the detrimental impact of snow and ice conditions which resulted in a cataclysmic decline in the number of caribou due to a reduction in forage availability (Mil ler and Gunn 2003). Large-scale climatic oscillations have been shown to explain a considerable proportion of annual variations in temperature and precipitation over large areas (Hurrell 1995; Broccol i 2001). Specifically, the Arctic Oscillation (AO) is closely related to the mean annual temperature recorded at different stations in the Arctic (Overpeck et al. 1997). The A O is a climate cycle and its index reflects the mean deviation from the average sea level pressure throughout the Northern Hemisphere at latitudes poleward o f 20°N (Thompson and Wallace 1998, 2001). The A O switches between two phases based on the deviation in mean pressure. In its positive phase, the Arctic experiences lower than average atmospheric pressure, which usually results in an increase in winter temperature and amounts of precipitation (Thompson et al. 2000). However, during the summer, the positive phase o f the A O usually results in cloudy conditions, decreased 39 temperatures and increased amount of precipitation. The negative phase of the A O corresponds to decreased winter temperatures and precipitation. The meteorological conditions of the negative phase during the summer are not well documented possibly because their impacts are not as pronounced. Recently, studies have demonstrated the indirect impacts o f large-scale climate oscillations on mammals and vegetation (Post and Stenseth 1999; Patterson and Power 2001; Aanes et al. 2002; Post and Forchhammer 2002; Hebblewhite 2005). In the Canadian Rockies, large-scale climate was influencing snow depth and winter climate, which resulted in greater difficulty for elk to avoid predation by wolves (Hebblewhite 2005). Other studies have shown that large-scale climate oscillations have influenced plant growth conditions, through changes in temperature and precipitation, which has then affected the population growth rate o f ungulates (Aanes et al. 2002). In one case, large-scale climate directly influenced plant phenology, as well as the reproduction and demography o f ungulates, and explained a greater amount of the variation than the local climate (Post and Stenseth 1999). There are direct and indirect influences of climate on caribou and the more information that is available on the impacts of climate, the better managers w i l l be able to interpret the complex dynamics o f this interaction. At present, the only annually-resolved climate data that are available for this area of the N . W . T . are from the distant meteorological station at Yellowknife, and this record is only consistently available back to 1943. Making inferences about climate-caribou relationships could be problematic when only one distant meteorological station is being used to study various caribou herds, occupying a vast geographic region. 40 Tree-rings have been used to reveal past climates at various spatial and temporal scales (Hughes 2002). Several studies have used tree-rings to reconstruct temperature in northern regions (Jacoby and Cook 1981; Jacoby et al. 1988; Jacoby and D 'Ar r igo 1989; Briffa et al. 1994). However, in the Northwest Territories dendroclimatological records are few and far between. The closest tree-ring chronologies to the sites in this study were developed from trees in the Franklin and Richardson Mountains, to the west of Great Bear Lake (Szeicz and MacDonald 1995). These sites are approximately 400km from the sites in this study, and in a different climatic and physiographic region. Although the mountainous regions o f the Northwest Territories are well studied, the dendroclimatological potential o f trees growing in the lowland areas o f the Northwest Territories to the southeast of Great Bear Lake has not been investigated. In an effort to increase the knowledge o f climate variability and its potential impacts on biological systems, as well as to expand on the number of tree-ring chronologies in this region, I present a dendroclimatic reconstruction of summer temperatures using a network o f tree-ring chronologies of black spruce (Picea mariana [Mil l . ] BSP) . The goal was to have a better understanding o f the local impact o f climate on the trees growing within the range o f the barren-ground caribou of the N . W . T . , more specifically at the same sites where caribou population abundance cycles were reconstructed in Chapter 2. In addition, I have included four tree-ring chronologies from the same region taken from the International Tree-ring Databank ( ITRDB) for comparison purposes. To explore the impact o f large-scale climate on tree growth, I compare the (AO) and tree growth in this region. 41 Specifically, this chapter aims to: (i) evaluate climatic factors correlated with radial growth o f Picea mariana along the latitudinal treeline in the central N . W . T , (ii) reconstruct summer climate in this region of the N . W . T . , and (iii) examine tree growth in relation to the Arct ic Oscillation. 3.2 M E T H O D S 3.2.1 Chronology development Seven ring-width chronologies were developed from samples taken from Picea mariana trees at treeline northwest and southeast o f Great Slave Lake. These sites were 7 o f the 19 sites selected to reconstruct caribou population abundance in Chapter 2. The sites were located along the treeline between latitudes of 61°- 65° N and longitudes of 106°-115° W , and were selected along the transect of 19 sites, to evenly represent the study area (Fig. 3.1, Table 3.1). The maximum distance between the two furthest sites is approximately 600 km. A l l sites were dominated by Picea mariana and had similar site characteristics. The sites were located within the forest-tundra with the landscape surrounded by broad upland and shallow lowland areas, rock outcrops, hummocky and ridged morainal deposits and eskers. In 2003 and 2004, a combination of both cores and cross-sections were collected from a minimum of 40 trees at each site. A n increment borer was used to extract two cores, at least 90° apart at breast height. Cross-sections were collected using a chainsaw, with all samples taken at breast height. Only healthy, undisturbed trees (i.e. without visual evidence of disturbance) were sampled. Cross-sections and cores were sanded using progressively finer grades of sandpaper (240, 320, 400 and 600 when necessary). The tree rings were visually and statistically crossdated to the calendar year of their formation by employing standard dendrochronological techniques as 42 125°W 120°W 1 1 5 ° W 6 5 ° N 60°N-13. y .8 Yellowknife _ Mac Kinky ,, t. ( h., 1 ~\ A A Great Slave L. 110°W 105°W Elevation (m) 10 - 1 0 0 rZlJ 101-300 701 -950 • Study sites - - - Treeline » Migratory route A ITRDB Aust in! 19. J 6 100 200 km I15°W I 110°W I 105°W K65°N F I G U R E 3.1 Map showing location of tree-ring chronology sites (numbered sites), sites from the International Tree-Ring Databank (ITRDB), Yellowknife meteorological station in the N . W . T . and the summer migratory route of barren-ground caribou. T A B L E 3.1 Site characteristics of Picea mariana chronologies along treeline in the N . W . T . Mean series Chronology Location 1 Time span length (years) No. trees No. radii Site 1 64° 27.2 N , 112° 45.6 W 1685-2003 154 34 66 Site 6 63° 32.7 N , 112° 18.8 W 1736-2003 126 33 64 Site 8 63° 13.9 N , 110° 55.2 W 1758-2003 151 34 66 Site 11 64° 47.7 N, 115° 12.2 W 1830-2003 92 31 58 Site 13 65° 09.8 N , 115° 37.5 W 1711-2003 158 34 68 Site 16 61° 45.1 N , 106° 29.0 W 1916-2004 60 39 78 Site 19 61° 36.0 N, 106° 49.3 W 1889-2004 68 35 69 Locations are shown in Figure 3.1 43 outlined by Stokes and Smiley (1968). Although there were few missing rings, cross-sections were preferred as they facilitated cross-dating. The width o f each dated ring was measured using a Quick Check QC-1000 connected to an AcuRite III digital encoder with a precision of 0.001 mm. These measurements and assigned dates were then verified using the computer program C O F E C H A (Holmes 1983). Once the crossdating errors were resolved using C O F E C H A and by re-examining the dates assigned to the samples, the computer program A R S T A N was used to develop the chronology. Ring-width measurements were standardized to remove growth-related trends while retaining the maximum common signal found in the tree-rings. Interactive detrending was used in A R S T A N to determine which curve was the best fit for each tree-ring series. A R S T A N plots the curve to the original data, and presents the resulting curve after each ring width is divided by the value of the growth curve. For the purposes o f this study, a negative exponential curve, a linear regression line of negative slope or a horizontal line through the mean was fitted to each series to remove the growth trends. The detrended indices from each year for all series are then averaged to obtain a master chronology. The residual chronology (instead of the Standard or Arstan chronologies) was used for each site because it was developed using autoregressive standardization techniques on the detrended ring-width series, which removes the effects of autocorrelation (Cook 1985). It is important to remove the effects o f autocorrelation in a dendroclimatic reconstruction since each tree-ring width must accurately represent the climate o f that particular year, and must not be an artifact o f a previous year's climate conditions. 44 To evaluate common features among tree-ring chronologies developed i n this study, a statistical analysis was carried out over the common period covered by all seven chronologies. Pearson's product moment correlations were computed to examine the between tree signal using data from the period 1916-2003. In order to provide additional information on the tree-growth response to climate in this region, I compared the chronologies developed in this study to four obtained from the I T R D B (Fig. 3.1) (Austin Lake, Mac Kin ley and Pethai Peninsula standard chronologies from Schweingruber, F . H . ; Coppermine standard chronology from Jacoby, G .C . , D 'Ar r igo , R .D . , Buckley, B . ; http://www.ncdc.noaa.gov/paleo/webmapper-treering.htmlv) using Pearson's correlations. The Pethai Peninsula and Austin Lake chronologies were derived from Picea glauca trees; the M a c Kin ley chronology was derived from P. mariana; and the Coppermine chronology was derived from samples of both Picea species. These I T R D B chronologies were selected based on their proximity to the tree-ring chronologies developed in this study, and because they were developed using only Picea species. 3.2.2 Multivariate Techniques Cluster analysis, a hierarchical clustering method that uses Euclidean distance as a measure o f dissimilarity, was used to examine the structure of the relationship among the residual chronologies. The tree-ring chronologies were used as variables, in this analysis. Principal components analysis ( P C A ) was performed to determine the explained variance among the residual chronologies. To evaluate the temporal stability of this shared variance, P C A s were performed for successive overlapping 30-year periods (1916-1946, 1926-1956, 1936-1966, 1946-1976, 1956-1986, and 1966-1996) and for the entire common period (1916-2003) (Tardif et al. 2002; Girardin and Tardif 2005). Overlapping 30-year periods were chosen since this is the 45 standard period (30-year climate normals) that is used by climatologists when analyzing climate patterns through time (http://climate.weatheroffice.ec.gc.ca/climate_normals/index_e.html). 3.2.3 Comparing meteorological data to the tree-ring chronologies The relationship between tree growth and climate was tested using Pearson's correlations and response function analysis (Fritts 1976), where monthly climate data are used as statistical predictors and ring-width chronologies as predictands (Davi et al. 2003). This analysis was performed using the P R E C O N program (Fritts 1999). The common time span (1943-2002) o f mean monthly temperature and total monthly precipitation and ring-width data from the seven chronologies were selected for this analysis. The growing period for trees near treeline in this region is expected to be between M a y and August, therefore the analysis was performed using meteorological data from the previous year's M a y to the current year's August for each chronology. The previous year's climate was included to account for the lag effect in tree growth. The lack of long-term meteorological stations in this region makes tree-growth-climate studies difficult. The closest station was used for this analysis (Yellowknife airport meteorological station (Fig. 3.1; 62°27'N, 114°26'W, 205.70 m a.s.l.; http://climate.weatheroffice.ec.gc.ca/prods_servs/cdcd_iso_e.html)). The data were truncated to the period between 1943-2002 to avoid missing data which were present prior to 1943. 3.2.4 Transfer functions To avoid problems of intercorrelation among the chronologies, a P C A was used to create a set of orthogonal (uncorrected) eigenvectors, which were then used as candidate predictor variables (Bhattacharyya and Chaudhary 2003; Girardin and Tardif 2005). Only the five longest 46 chronologies were used in order to obtain the longest dendroclimatic reconstruction possible (Table 3.1). If all seven chronologies had been used, the reconstruction would only extend to 1916 since this is the length of the shortest chronology, which is very short by dendroclimatic standards. For the purpose of this study, the chronology needed to extend to at least 1900 to provide a long-enough time-frame for comparison to the caribou abundance cycles. The eigenvectors were then used to represent the signal derived from five chronologies. Based on the results of the response function analysis, transfer function models predicting summer temperatures from five eigenvectors were calibrated using stepwise multiple linear regression over the common period of analysis (1943-2002). A calibration was done for preliminary models to predict temperature for individual months of the growing season (May-September), and for different combinations of months (Case and MacDonald 1995; Rayback and Henry 2006). Forward (t +1) and backward (t - 1) lagged indices were added to capture persistence effects in the growth-climate relationships (Fritts 1976; Jacoby et al. 1985; Briffa et al. 1988). A data-splitting technique was used to examine the temporal stability o f the derived model (Fritts 1976). The entire data set (1943-2002) was split in two halves: the "early" period (1943-1972) and the "late" period (1973-2002). Following this, a transfer function was calibrated for each period using stepwise regression. Each reconstruction was then verified with the opposite period; for example, the early model was verified on the 1973-2002 period and the late model was verified on the 1943-1972 period. Verification statistics included Pearson's product-moment correlation coefficient (r), reduction of error statistic (RE; positive values demonstrate skill in reconstruction), the coefficient of efficiency (CE; positive values demonstrate skill in reconstruction), and the nonparametric sign test of first differences (Briffa et al. 1988). 47 3.2.5 The Arctic Oscillation (AO) The A O index is primarily evaluated and modeled during winter months (Shindell et al. 1999; Hartmann et al. 2000). However, recent reductions in atmospheric pressure have been observed during summer months (AOS) , reflecting the same nature as the A O winter index ( A O W ) (Serreze et al. 1997, 2000; Aanes et al. 2002; D 'Ar r igo et al. 2002). This may indicate that the A O can have impacts on weather patterns during the summer season. The A O W index is based on the months October-May, while the A O S index reflects the plant growth season from June-August, corresponding to the snow-free months in the study area. The phases of the A O S have not been well documented, therefore I developed a method to distinguish these phase changes through time. I performed a 5 t h order regression (y = 5E-07x 5 -O.OOOlx4 + 0.0121x 3 - 0.4543x 2 + 4.9464x + 34.282; R 2 = 0.29) with the A O S index as the dependent variable and year as the independent variable. The change in the slope o f the regression was then used to determine shifts in the A O S . Based on the analysis of the changes in the slope, the A O S index has been divided in the following four time periods: 1900-1930, 1931-1955, 1956-1983 and 1984-2000 (Fig. 3.2). 3.2.6 Relationships between the Arctic Oscillation (AO) and tree-ring chronologies To establish whether there was a link between large-scale climate (AO) and tree-growth, a response function analysis was computed between all seven chronologies and the monthly index o f the A O for the entire period of analysis (1900-2000). Spearman's rank correlations were computed for the four phases o f the A O (1900-1930, 1931-1955, 1956-1983 and 1984-2000), to determine whether the chronologies are more responsive to certain phases of the A O . Pearson's 48 correlations were used to analyze the relationship between negative (<zero) A O index years and the tree-ring chronologies. The same analysis was performed using only positive (>zero) A O index years. This correlation analysis was used to determine whether tree growth is responding to both phases of the A O (positive and negative) or i f only specific phases of the A O are affecting tree growth. c - I . O 1900 1920 1940 1960 1980 2000 Year F I G U R E 3.2 The standardized Arct ic Oscillation (AO) anomaly index for June to August for the period between 1900-2002, and a 5 t h order regression curve (data from http://www.jisao.washington.edU/ao/#monthly). The values of the A O S have been multiplied by -1 to illustrate the positive and negative phases. 3.3 R E S U L T S 3.3.1 Descriptive statistics and trends of chronologies The longest chronology was 319 years and covered the period 1685-2003 and the shortest chronology was 88 years and covered the period 1916-2004 (Fig. 3.3; Table 3.1). Descriptive statistics such as standard deviation (s.d.), mean sensitivity, and first order autocorrelation (r l ) , have been included for each chronology to allow comparisons with other studies (Table 3.2). 49 X T3 M $ s - r—i 1700 1750 1800 1850 Year 1900 1950 200Q F I G U R E 3.3 Ring-width chronologies from the seven sampled sites (ring-width sample depth is below each chronology curve). The thick line is a five-year moving average. 50 T A B L E 3.2 Descriptive statistics for the seven chronologies developed using Picea mariana. s.d. = standard deviation; r l = first-order autocorrelation; S N R = signal-to-noise ratio; V A R p c l = variance explained by the first component; EPS = expressed population signal. Common interval Standard Chronology Residual chronology Detrended series Chronology Mean sensitivity s.d. r l Mean Sensitivity s.d. Common interval SNR VARpcl EPS Site 1 0.17 0.32 0.78 0.19 0.17 1892-2003 16.00 31.0% 0.94 Site 6 0.18 0.28 0.80 0.17 0.14 1921-2003 8.61 27.0% 0.90 Site 8 0.16 0.41 0.87 0.18 0.17 1900-2003 21.26 32.8% 0.96 Site 11 0.19 0.38 0.91 0.15 0.14 1940-2003 17.58 39.6% 0.95 Site 13 0.16 0.35 0.83 0.17 0.18 1890-2003 18.57 . 33.2% 0.95 Site 16 0.15 0.23 0.68 0.17 0.13 1952-2004 11.22 23.0% 0.92 Site 19 0.13 0.18 0.66 0.14 0.12 1948-2004 11.03 28.4% 0.92 Tree-ring chronologies with high s.d., high mean sensitivity and low autocorrelation are considered to be the most suitable for dendroclimatic reconstructions (Fritts 1976). The signal to noise ratio (SNR) was used to determine the strength o f the signal between trees (common variance) (Wigley et al. 1984; Briffa and Jones 1990). The S N R of the chronologies ranged between 8.61 and 21.26. The variance explained by the first component ( V A R p c l ) indicates the variance represented by the common forcing factor at each site, which could be a climatic signal experienced by all trees in a region. A l l seven chronologies had similar common variance with values ranging from 23% to 39.6% (Table 3.2). Expressed Population Signal (EPS) is used as a measure o f the strength of the common signal within each chronology (Wigley et al. 1984). The average EPS for all seven chronologies was 0.93 (all are above 0.90), which is well above the accepted threshold value of 0.85 (Table 3.2; Wigley et al. 1984). A l l seven chronologies depicted similar trends during the period of analysis (1916-2003; Fig . 3.3). For those chronologies that include the 1800s (all except Sites 11, 16 and 19), ring-widths 51 were narrow, with less variability, during the early 1800s until approximately 1900. Growth rates were greater in the early 20 t h century compared to the 1800s, however trends remained relatively stable throughout the rest of the century. T A B L E 3.3 Pearson's correlations among the seven residual chronologies developed from samples taken at treeline, N . W . T . A l l correlations were significant at p<0.05; n = 88. Site 6 Site 1 Site 8 Site 11 Site 13 Site 16 Site 1 0.77 Site 8 0.78 0.75 Site 11 0.61 0.49 0.67 Site 13 0.72 0.56 0.71 0.75 Site 16 0.38 0.35 0.46 0.34 0.39 Site 19 0.43 0.49 0.56 0.36 0.35 0.47 A l l o f the pairs of correlations between residual chronologies were significant and positive, with the mean correlation coefficient between pairs of sites being 0.54 (Table 3.3). The highest correlation occurred between Site 8 and Site 1 (0.78). Sites 16 and 19 showed the lowest correlations which may be due to their distance from the other sites. The fact that all chronologies were highly correlated was especially remarkable considering that the two furthest sites were approximately 600 km apart. Correlations among the seven chronologies developed in this study and the four obtained from the I T R D B ranged between -0.07 and 0.49 (Table 3.4). The Mac Kin ley chronology was highly correlated to all other chronologies except for Site 16. Mac Kinley was the only I T R D B chronology that was correlated to all 3 other I T R D B chronologies. Pethai and Austin were correlated to Sites 1, 6 and to each other. Pethai was also correlated to the Mac Kinley chronology. Coppermine was correlated to all o f the chronologies developed in this study, except Sites 11 and 19. 52 T A B L E 3.4 Pearson's correlations among the seven residual treeline chronologies and the I T R D B chronologies (Aus. = Austin; Copp. = Coppermine; Mac. = Mac Kinley; Pet. = Pethai.). Marked correlations are significant at p<0.05, n = 62. Site 1 Site 6 Site 8 Site 11 Site 13 Site 16 Site 19 Aus. Site 6 0.78 Site 8 0.74 0.73 Site 11 0.52 0.49 0.63 Site 13 0.70 0.65 0.71 0.63 Site 16 0.29 0.33 0.41 0.25 0.31 Site 19 0.38 0.46 0.52 0.36 0.39 0.42 Aus. 0.37 0.30 0.18 0.23 0.17 0.10 0.21 Copp. 0.29 0.31 0.34 0.25 0.34 0.25 0.05 -0.02 Mac. 0.49 0.47 0.49 0.39 0.50 0.04 0.28 0.30 Pet. 0.40 0.42 0.30 0.28 0.26 -0.07 0.13 0.42 Copp. Mac. 0.32 0.17 0.44 Since Coppermine was the furthest site from the other three I T R D B chronologies, it was the least correlated among the I T R D B chronologies, however it was correlated to most o f the other chronologies. The analysis of the correlations and the trends in the chronologies demonstrated that there was a significant amount of variance that was common among most sites used in this study. 3.3.2 Mul t ivar ia te analysis Using cluster analysis I identified two clusters (Fig. 3.4). Sites 16 and 19 formed their own cluster, which was likely due to the fact that they were located further from the other chronologies (Fig. 3.1). The remaining sites formed clusters with the sites to which they were closest in proximity, for example, Sites 1 and 6 formed a cluster, followed by Sites 8, 11 and 13. Based on the P C A of the seven chronologies for the common time period 1916-2003, the cumulative explained variance o f the first two principal components was 74.9% (Figure 3.5). The first component explaind 61.8% of the variance, which demonstrates the similarity among the 53 chronologies, and it was correlated with all seven chronologies. The second component (explained variance = 13.06%) was positively correlated to site 11 and 13, but highly negatively correlated to site 16 and 19 and not correlated to sites 1, 6 and 8. Site 1 Site 6 Site 8 Site 11 Site .1.3 Site 16 Site 19 0.95 1.00 1.05 1.10 . 1.15 1.20 1.25 Linkage Distance F I G U R E 3.4 Hierarchical cluster analysis of seven residual chronologies. 1.0 0.5 S i t c J : I 0 S i t e l l • .Site.1 S i t e o 8 * s i * e 6 -1.0 -0.5 0.5 1.0 Site 19 • • Site 16 -0.5 -1.0 PC I PC 2 F I G U R E 3.5 Plot of the loadings for the first two principal components for each chronology. The percentage of explained variance for P C I and PC2 is 61.80% and 13.10%, respectively. 54 The results from the P C A s calculated for successive overlapping 30-yr periods indicated that the percentage o f explained variance by the two first principal components was relatively constant through time (Fig. 3.6). The variance of the first component peaked in 1936-1966, after which period it remained generally constant. The variance explained by the second principal component remained relatively constant, with a slight low between 1926-1956. 1916- 1926- 1936- 1946- 1956- 1966-1946 1956' 1966 1976 1986 1996 Time span FIGURE 3.6 Percent variance expressed by the first and second principal components of P C A derived from the seven residual chronologies for the subperiods: 1916-1946, 1926-1956, 1936-1966, 1946-1976, 1956-1986, and 1966-1996. For the whole common period 1916-2003, the variance explained by the first and second principal components is 74.9%. 3.3.3 Relationship between meteorological data and tree-ring chronologies Response function analyses indicated that at all sites, except Site 19, there was a significant relationship between Picea mariana growth and July temperatures of the current year (Fig. 3.7) Tree growth at Site 19 was positively correlated with July temperatures, but the correlation was not significant. August temperatures were consistently, positively correlated to ring-width, however only one of the chronologies had significant coefficient values. In addition, there was a 55 significant negative correlation to July temperatures of the previous year at all sites, except for Site 19. Mean temperature Total precipitation Site 1 M. J i' A S O N D J F M A M T 1 A M J J A S O N D J F M A M J J A •Site 6 0.4 f 0.2 0.0 -0.2 m l E X C3 M J .1 A S O N D. J F U A :M J. J A M 1 J A S O N D J F M A M j ; 'j A cH rta _ _ Site 8 0.4 CD a 0.0 im I I MJ J A S O N D J F M A 'M .1 J A M J J A. S O N D J F M A M J J A 0 . 4 , Site 11 0.2 0.0 -0.2 0.4 0.2 0.0 ^0.2 J u M J J A S 0 N D J F M A M J .1 A previous year | growth year M J .1 A S O N D I F M A M J J A previous year growth year 56 0.4 0.2-1 Mean temperature Total precipitation Site 13 L fl o.o o -0.2 a P M J J A S O N D j F M A M 1 J A 0.4 0:2 0.0 -0.2 J Z L fl 0.4 O fl 3 0-2 o & Pi -0. 0.0 -1 1 ^ Site 16 M J J A S O N D J F M A M J j A -0.2 M J .1 A S 0 N D J F M A M .1 j A . M J J A, S 0 N D J F M A M J J A o.41 — , Site 1.9 0.4 0.2 0.0 -02 M J J A S O N D J F M A M J J A previous year | growth year M J J A S O N D J F M A M .1 .1 A previous year | growth.year F I G U R E 3.7 Response function coefficients relating mean monthly temperature and total monthly precipitation to ring-width chronologies at treeline sites, N . W . T . Shaded coefficients are significant at p<0.05. There was also a general tendency for August and September temperatures of the previous . growth year to be negatively related to tree-ring growth. Interestingly, October temperatures o f the previous year were positively related to ring-widths in five o f the seven sites. There was poor correlation at all sites between tree-ring growth and total monthly precipitation. There was a 57 tendency for June precipitation of the previous growth year to be positively correlated to tree-ring growth, although the correlation was significant only at Site 6. Pearson correlations were also performed for mean monthly temperature (Appendix A : Table A.2) and total monthly precipitation (Appendix A : Table A.3) for the same time period. These results agreed with those yielded by the response function analysis. July and August temperatures were most significantly correlated to tree-growth and there was less (non-significant) correlation with total monthly precipitation. 3.3.4 Summer temperature reconstruction Based on the results of the response function analyses, mean July to August temperature (averaged for both months) of the growing season were selected as the variable to be reconstructed. July to August temperatures were used rather than only July temperatures as the inclusion o f August temperatures helped explain a greater amount of variance in the final model. The P C A o f the five longest chronologies yielded five eigenvectors (Factors) that were then used to represent the chronologies. The results of the stepwise multiple linear regression resulted in the maximum variance (40%) being explained by the inclusion of Factor 1 , F a c t o r 4h and Factor \ t . Factor 1 i s the first factor lagged by 1 year, and Factor 4 and Factor 1 were of the current growth year. This percentage of explained variance is comparable to other reconstructions (Watson and Luckman 2004). Early (1943-1972) and late (1973-2002) calibration models accounted for 58% and 39% of the variance in the July-August temperature data, respectively. For the verification o f the early (1943-1972) and late (1973-2002) periods, the 58 Pearson's correlation coefficients (r = 0.61 and r = 0.50) were significant (Table 3.5), which renders the model adequate for estimating values not contained in the calibration period. T A B L E 3.5 Calibration and verification statistics for predicting July-August temperatures from eigenvectors derived from five chronologies at treeline in the N . W . T . Calibration Verification r r 2 adjr n r ST RE C E 1943-1972 0.764 0.584 0.517 1973-2002 30 0.501 20/9 0.430 0.370 1973-2002 0.621 0.386 0.341 1943-1972 30 0.607 21/8 0.487 0.405 Full model 1943-2002 July-Aug temp 0.635 0.403 = 15.264+ (0.426 * F 1 , 0.371 . ) - (0.282 * F4 t) -(0.186 * F l t ) n = sample size F = Factor ST (sign test) = number of agreements/disagreements; nonparametric sign test of first differences (Fritts 1976) R E = reduction of error (Fritts 1976) C E = coefficient of efficiency Four independent verification tests demonstrated that the climatic information in the reconstruction is adequate to apply it beyond the calibration period (Table 3.5). The reduction of error (RE) (0.43; 0.49) and coefficient of efficiency (CE) (0.37; 0.41) values were positive which indicates that the model is stable and has predictive capabilities (Briffa et. al 1988). The reconstruction performed well in the sign test as there were a significantly larger number of agreements than disagreements, indicating that there was sufficient similarity existing between the actual and estimated data (Fritts 1976). The model passed the verification tests and can therefore be used to estimate July-August temperatures (Fritts 1976). Thus, the full model calibrated over the entire time period (1943-2002) was considered statistically stable. The model was used to reconstruct summer (July-August) temperatures to 1831 (Fig. 3.8). The record was restricted to the longest of the five chronologies chosen for this analysis (1831-2002) and by lags included in the model. The time series of the observed and estimated reconstruction for July-August temperatures were significantly correlated (r 2 = 0.63). 59 • a i l j r\ >v» • h iA>A A/w iuU ,k i k i l i 1 lAIlt hfifl III i I V S i 3T ill ^ ft ' i .» P 11 1850 1900 1,950 2000 Year F I G U R E 3.8 Observed (dashed line) and reconstructed (solid line) July-August temperatures at treeline, N . W . T . Darker solid line is a 3-year running mean o f the reconstructed July-August temperatures. Reconstruction covers the time span 1831-2002. In general, the reconstructed July-August temperatures followed closely the trends found in the observed meteorological data. However, the reconstruction tended to underestimate the warmest and coldest years (1947, 1950, 1952, 1954, 1968, 1978, 1981, 1985, 1989, 1994, and 1998) and was marked by lower variability in the cold years in comparison to the observed time series. Despite these differences, the reconstruction demonstrated below average temperature values until approximately 1900, followed by a slow increase in temperature until the early 1940s, after which summer temperatures became increasingly variable. 3.3.5 Comparing tree-ring chronologies to the AO During the last century, the A O S has undergone four phases: a positive phase from 1900-1930, a negative phase from 1931-1955, a positive phase from 1956-1983 and a negative phase from 1984-2000 (Fig. 3.9; http://tao.atmos.washington.edu/ao/#monthlv). The first two phases of the A O S have a clear pattern whereby a high index is seen during the positive phase and a low index during the negative phase. However, the strength of the last two phases (1956-1983 and 1984-2000) is less evident with the index showing more episodic and variable behavior. 60 Arctic Oscillation (JJAH 1900 1920 1940 1960 1980 Year F I G U R E 3.9 The standardized Arct ic Oscillation (AO) anomaly index for June to August for the period between 1900-2002, and a 5 t h order regression curve (data from http://tao.atmos.washington.edU/ao/#monthly). The values o f the A O S have been multiplied by -1 to illustrate the positive and negative phases. The response function analysis computed between the tree-ring chronologies and the monthly index o f the A O demonstrated that all sites except Sites 1 and 6 were positively correlated to July index o f the A O (Appendix A ; Table A.4) . Only Site 11 had a significant positive correlation to June index of the A O . A l l sites except Site 16 were correlated to the previous November index of the A O . Since the July index of the A O had the highest correlations to tree-ring growth, I used this month to determine whether the different phases of the A O were correlated to tree growth. The chronologies showed no correlations to the first two phases of the July A O (1900-1930 and 1931-1955), however Sites 8, 11 and 16 were correlated to the July A O for the period from 1956-1983 (Appendix A ; Table A.5) . None o f the sites were correlated to the July index o f the A O for the period from 1984-2000. 61 3.4 D I S C U S S I O N The seven chronologies derived from Picea mariana provided proxy climate data in a region where climate records are few and short in length. A l l seven chronologies depicted similar trends over the common time period. The fact that the chronologies are all correlated to each other and that there is a significant amount of correlation to the distantly located I T R D B sites, indicates that common climatic factors are affecting the growth o f these trees in this area. The longer chronologies all demonstrated low growth rates during the earlier part o f the time series until approximately 1900, which suggests that the tree-rings were recording the cooling near the end of the Little Ice Age (Gajewski and Atkinson 2003). Cluster analysis and P C A confirmed that the tree-ring chronologies developed in this study were experiencing similar trends across all sites. The factors most highly correlated to ring-width were July-August mean temperatures, which were used in the dendroclimatic reconstruction. It has been generally found that growth o f trees in northern regions is mainly limited by summer temperatures (Tranquillini 1979; Black and Bliss 1980). The geographic position of the treeline, where trees grow at the limit o f their ecological tolerance, is controlled by climate, and is associated with the position of the 10°C mean July isotherm in North America (Sirois 1992). Its geographic position is similar to the summer position o f the arctic front (Beringer et al. 2001), which is the zone o f transition between dry, cold arctic air masses and warm, moist air masses from the south. Therefore, it is not surprising that July temperatures influence tree growth in this region. One reason for the lack o f correlation between total monthly precipitation and tree-growth is that precipitation is an intermittent phenomenon and is more regionally variable than average temperature. In addition, the short climate record and the long distance to the nearest climate station (Yellowknife) 62 contribute to the low correlations observed in the response function. Tree-growth could be limited by precipitation in this region but the aforementioned scale-dependent factors affect the results of the response function analysis. This study provided the first dendroclimatic reconstruction of summer temperatures in this region of central N . W . T . The reconstruction was correlated to the observed meteorological data, however, it tended to underestimate variability in the climate data. It is possible that the effect o f variability in temperature which was recorded in the instrumental data was only partly captured in the tree-ring record since the effects of extreme temperatures is often reflected in tree rings over two consecutive years. Despite the underestimation of extreme temperatures, the reconstruction captured long-term trends including the below average temperatures until approximately 1900, followed by a slow increase in temperature until the early 1930s, after which summer temperatures became increasingly variable until the end of the record. The lower temperatures in the early part of the reconstruction correspond with the end of Little Ice Age, where temperatures were lower than average (Cropper and Fritts 1981; Szeicz and Macdonald 1995; Ruddiman 2001; Davi et al. 2003). Although the reconstruction demonstrated lower variability in the common period (1943-2002) compared to the meteorological record, this period was generally more variable than the rest of the reconstruction. There was a strong relationship between the July index of the A O and tree-growth at most sites, but the relationship has not remained constant through time. Tree-ring chronologies were not correlated to the first two phases o f the A O , however the positive phase from 1956-1983 had an influence on tree growth at 3 o f the 7 sites. The change in the response o f tree-growth to the A O 63 after 1950 corresponds well with the increased variability as was seen in both the instrumental data and the July-August reconstruction (Fig. 3.8). A similar variability in both tree-ring chronologies and summer temperature reconstructions was found in the Northwestern region of the N . W . T . (Szeicz and Macdonald 1995). In summary, summer temperatures strongly affected ring-width growth i n Picea mariana at the sites in the central N W T . Both the tree-ring chronologies and the temperature reconstruction demonstrated similar trends to other chronologies developed in the Northwestern region o f the N . W . T . (Jacoby et al. 1985; Szeicz and MacDonald 1995). In Chapter 2,1 demonstrated that two groups of caribou herds were experiencing similar population abundance cycles since 1900. These simultaneous changes in abundance and the similar patterns in ring-width chronologies are an indication that the region is influenced by a major climate signal. The next objective is to investigate the relationship between long-term caribou abundance and the A O . 64 3.5 REFERENCES Aanes, R., Saether, B . , Smith, F . M . , Cooper, E.J . , Wookey, P .A . , and 0ritsland, N . A . 2002. The Arctic Oscillation predicts effects of climate change in two trophic levels in a high-arctic ecosystem. Ecology Letters, 5: 445-453. Beringer, J., Tapper, N . J . , McHugh, I., Chapin III, F.S, Lynch, A . H . , Serreze, M . C . , and Slater, A . 2001. Impact o f Arct ic treeline on synoptic climate. Geophysical Research Letters, 28: 4247-4250. Bhattacharyya, A . and Chaudhary, V . 2003. Late-Summer Temperature Reconstruction o f the Eastern Himalayan Region Based on Tree-Ring Data o f Abies densa. Arctic, Antarctic, and Alpine Research, 35: 196-202. Black, R . A . , and Bliss , L . C . 1980. Reproductive ecology of Picea mariana (Mi l l . ) BSP . , at treeline near Inuvik, Northwest Territories, Canada. Ecological Monographs, 50: 331-354. Briffa, K . R., Jones, P. D . , Pilcher, J. R., and Hughes, M . K . 1988. Reconstructing summer temperatures in northern Fennoscandinavia back to A . D . 1700 using tree-ring data from Scots pine. Arctic and Alpine Research, 20: 385-394. Briffa, K . R. and Jones, P. D . 1990. Basic chronology statistics and assessment. In: Cook, E . and Kairiukstis, L . (eds.), Methods of Dendrochronology: Applications in the Environmental Sciences. Dordrecht: Kluwer Academic Publishers. Briffa, K . R . , Jones, P .D. , and Schweingruber, F . H . 1994. Summer temperatures across northern North America: regional reconstructions from 1760 using tree-ring densities. Journal of Geophysical Research, 99: 25835-25844. 65 Broccoli , A . J . , Delworth, T .L . , and Lau, N . - G . 2001. The effect of changes in observational coverage on the association between surface temperature and the Arct ic Oscillation. Journal of Climate, 14: 2481-2485. Case, R. A . , and MacDonald, G . M . , 1995. A dendroclimatic reconstruction of annual precipitation on the western Canadian prairies since A . D . 1505 from Pinus flexilis James. Quaternary Research, 44: 267-275. Cook, E .R . 1985. A time series analysis approach to tree-ring standardization. Dissertation, University of Arizona. Cropper, J. P. and Fritts, H . C. 1981. Tree-ring width chronologies from the North American Arctic, Arctic and Alpine Research, 13: 245-260. D 'Ar r igo , R .D . , Cook, E.R. , Mann, M . E . , and Jacoby, G .C . 2002. Tree-ring reconstructions of temperature and sea-level pressure variability associated with the warm-season Arct ic Oscillation since A D 1650. Geophysical Research Letters, 30: 3-1. Davi , N . K . , Jacoby, G . C , and Wiles, G .C . 2003. Boreal temperature variability inferred from maximum latewood density and tree-ring width data, Wrangell Mountain region, Alaska. Quaternary Research, 60: 252-262. Fritts, H . C 1999. P R E C O N 5.17b User's Manual, dendrochronological modeling. Tuscon, A Z . Fritts, H . C . 1976. Tree Rings and Climate. New York : Academic Press, 567 pp. Gajewski, K . and Atkinson, D . A . 2003. Climatic change in northern Canada. Environmental Reviews, 11: 69-102. Girardin, M . - P . , and Tardif, J. 2005. Sensitivity o f tree growth to the atmospheric vertical profile in the Boreal Plains of Manitoba, Canada. Canadian Journal of Forest Research, 35: 48-64. 66 Hartmann, D . L . , Wallace, J . M . , Limpasuvan, V . , Thompson, D .W.J . , and Holton, J.R. 2000. Can ozone depletion and global warming interact to produce rapid climate change? Proceedings of the National Academy of Sciences of the United States of America, 97: 1412-1417. Hebblewhite, M . 2005. Predation by wolves interacts with the North Pacific Oscillation (NPO) on a western North American elk population. Journal of Animal Ecology, 74: 226-233. Holmes, R. L . 1983. Computer assisted quality control in tree ring dating and measuring. Tree Ring Bulletin, 43: 69-78. Hughes, M . K . 2002. Dendrochronology in climatology - the state of the art. Dendrochronologia, 20:95-116. Hurrell, J .W. 1995. Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation. Science, 269: 676-679. Jacoby. G . C , and Cook, E . R. 1981. Past temperature variations inferred from a 400-year tree-ring chronology from Y u k o n Territory, Canada. Arctic and Alpine Research, 13: 409-418. Jacoby, G .C . , Cook, E .R . , and Ulan, L . D . 1985. Reconstructed summer degree days in central Alaska and northwestern Canada since 1524. Quaternary Research, 23: 18-26. Jacoby, G .C . , Ivanciu, I.S., and Ulan, L . D . 1988. A 263-year record of summer temperature for northern Quebec reconstructed from tree-ring data and evidence of a major climatic shift in the early 1800's. Palaeogeography, Palaeoclimnology, Palaeoecology, 64: 69-78. Jacoby, G .C . , and D 'Ar r igo , R. 1989. Reconstructed Northern Hemisphere annual temperature since 1671 based on high-latitude tree-ring data from North America. Climatic Change, 14: 39-59. 67 Miller , F . L . , and Gunn, A . 2003. Catastrophic die-off o f Peary caribou on the Western Queen Elizabeth Islands, Canadian High Arctic. Arctic, 56: 381-390. Overpeck, J. , Hughen, K . , Hardy, D . , Bradley, R., Case, R., Douglas, M . , Finney, B . , Gajewski, K . , Jacoby, G . , Jennings, A . , Lamoureux, S., Lasca, A . , MacDonald, G . , Moore, J., Retelle, M . , Smith, S., Wolfe, A . , and Zielinski, G . 1997. Arct ic environmental change of the last four centuries. Science, 278: 1251-1256. Patterson, B .R . , and Power, V . A . 2001. Contributions o f forage competition, harvest, and climate fluctuation to changes in population growth of northern white-tailed deer. Oecologia, 128: 244-255. Post, E , and Forchhammer, M . C . 2002. Synchronization of animal population dynamics by large-scale climate. Nature, 42: 168-171. Post, E . , and Stenseth, N . C . 1999. Climatic variability, plant phenology, and northern ungulates. Ecology, 80: 1322-1339. Ranta, E . , Kaitala, V . , and Lundberg, P. 1997. The spatial dimension in population fluctuations. Science, 278: 1621-1623. Rayback, S.A., and Henry, G .H .R . 2006. Reconstruction o f summer temperature for. a Canadian High Arctic site from retrospective analysis o f the dwarf-shrub, Cassiope tetragona. Arctic, Antarctic, and Alpine Research, 38: 228-238. Ruddiman, W . F . 2001. The Earth's Climate: Past and Future. New York : W . H . Freeman and Company. Serreze, M . C , Carse, F. , Barry, R . G . , and Rogers, J . C 1997. Icelandic low cyclone activity: Climatological features, linkages with the N A O , and relationships with recent changes in the Northern Hemisphere circulation. Journal of Climate, 10: 453-464. 68 Serreze, M . C . , Walsh, J.E., Chapin, F.S. I l l , Osterkamp, T., Dyurgerov, M . , Romanovsky, V . , Oechel, W . C . , Morison, J. , Zhang, T., and Barry, R . G . 2000. Observational evidence of recent change in the northern high-latitude environment. Climatic Change, 46: 159-207. Shindell, D.T. , Mi l le r , R . L . , Schmidt, G . A . , and Pandolfo, L . 1999. Simulation of recent northern winter climate trends by greenhouse-gas forcing. Nature, 399: 452-455. Sirois, L . 1992. The transition between boreal forest and tundra. Eds. H . H . Leemans and G . B . Bonan. A Systems Analysis of the Global Boreal Forest. Cambridge University Press, pp. 196-215. Stokes, M . A . , and Smiley, T . L . 1968. An introduction to tree ring dating. University of Chicago Press: Chicago. Szeicz, J. M . and G . M . MacDonald. 1995. Recent white spruce dynamics at the subarctic alpine treeline of north-western Canada. Journal of Ecology, 83: 873-885. Tardif, J., Conciatori, F., and Bergeron, Y . 2002. Comparative analysis o f the climatic response o f seven boreal tree species from northwestern Quebec, Canada. Tree-Ring Research, 57: 25-37. Telfer, E.S., and Kelsal l , J.P. 1984. Adaptation of some large North American Mammals for survival in snow. Ecology, 65: 1828-1834. Thompson, D .W.J . , and Wallace, J . M . 1998. The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophysical Research Letters, 25: 1297-1300. Thompson, D . W . J., Wallace, J. M . , and Hegerl, G . C. 2000. Annular modes in the extratropical circulation, Part II: Trends. Journal of Climate, 13: 1018-1036. Thompson, D .W.J . , and Wallace, J . M . 2001. Regional climate impacts of the Northern Hemisphere annular mode. Science, 293: 85-89. 69 Tranquillini, W . 1979. Physiological ecology of the alpine timberline. Berl in: Springer. Watson, E . , and Luckman, B . H . 2004. Tree-ring based reconstructions o f precipitation for the southern Canadian Cordillera. Climatic Change, 65: 20-241. Wigley, T. M . L . , Briffa, K . R., and Jones, P. D . , 1984. On the average value o f correlated time series, with applications in dendroclimatology and hydrometeorology. Journal of Climate and Applied Meteorology, 23: 201-221. 70 CHAPTER 4. THE RELATION BETWEEN LONG-TERM ABUNDANCE CYCLES IN BARREN-GROUND CARIBOU AND THE ARCTIC OSCILLATION 4.1 INTRODUCTION Climate is one o f the most influential factors driving the population dynamics o f ungulates through various direct and indirect effects on forage quality and availability, predation, and abundance o f biting insects. In the winter, deep snow can prevent caribou from finding highly digestible lichen (Turner et al. 1994). The quality and availability of forage directly influences female caribou body condition, which then affects calf survival (Skogland 1985; Portier et al. 1998). Conversely, a year with reduced snow pack can help decrease predation rates since the caribou are able to run through the snow easier, thereby avoiding predation (Telfer and Kelsal l 1984; Nelson and Mech 1986; Huggard 1993; Post et al. 1999; Hebblewhite et al. 2002; Hebblewhite 2005). The influence of mosquitoes can increase dramatically during wet, and cloudy summers (Downes et al. 1986; Noel et al. 1998), which can result in lower calf survival and reduced caribou numbers. Recent studies reveal widespread effects of large-scale climatic patterns or oscillations on the population dynamics of ungulates (Post and Stenseth 1999; Patterson and Power 2001; Aanes et al. 2002; Griffith et al. 2002; Post and Forchhammer 2002; Hebblewhite 2005). The index of the Arctic Oscillation (AO), a large-scale climate oscillation, is based on the mean deviation from the average sea level pressure throughout the Northern Hemisphere at latitudes poleward of 20°N (Thompson and Wallace 2001). The A O has been shown to explain a considerable portion o f the variation in mean annual temperatures recorded at different stations in the Arctic (Chapter 3; Overpeck et al. 1997). Several studies have demonstrated the indirect impacts of such large-scale 71 climate oscillations on the population dynamics of different mammals. For example, positive values o f the A O were associated with a decline in reindeer population growth rate in Svalbard (Aanes et al. 2002). The positive phase of the North Pacific Oscillation (NPO) had an indirect effect on elk population growth rate by increasing winter severity in Banff National Park (Hebblewhite 2005). A similar indirect effect of the positive phase of the North Atlantic Oscillation ( N A O ) was experienced with population abundances of muskoxen and caribou, which were affected by changes in winter climate (Forchhammer et al. 2002). A recent synthesis o f research has shown that the Arctic is experiencing unprecedented increases in temperature ( A C I A 2004). Climate change predictions have implied that caribou/reindeer are l ikely to be stressed as their access to food sources, breeding grounds and migration routes are altered ( A C I A 2004). One of the major challenges for biologists is to predict the future impact o f global warming on animal populations. However, to differentiate natural variation from the impacts of global warming, we must first understand how climate has impacted caribou abundance in the past. A t present, the data describing caribou population abundance is limited to the last three or four decades when relatively accurate estimates o f population size have been available from aerial surveys (Caughley and Gunn 1993; Gunn 2005). In addition, qualitative data on the population dynamics o f caribou since at least the 1920s has been collected through narratives of aboriginal community members in the form of traditional knowledge (Dogrib Treaty 11 Council 2001). 72 A lack o f annually resolved archival data on demographic trends for the past century makes it difficult to support any hypotheses about the causal factors associated with changes in caribou abundance through time. Recently, a new method was developed to reconstruct past caribou abundance cycles using dendroecology (Morneau and Payette 1998, 2000; Boudreau et al. 2003; Zalatan et al. 2006). Trampling scars formed by caribou on superficial roots and low branches remove part of the bark, leaving a scar. The scars are then accurately dated and their changes in frequency used as a proxy for caribou abundance through time. Dating a number o f scars distributed along caribou trails over the herd's annual range can provide an index of caribou activity over a longer time period than other methods. Using a targeted approach to find the oldest scars possible is essential to develop a long proxy record o f caribou abundance (Chapter 2). However, the nature o f how the scars are formed on the roots results in a dataset that is grouped into 5-year age-classes. Although this method is limited because it does not provide annually resolved estimates of caribou abundance, it does provide a much longer record o f caribou abundance cycles than other methods. This long term record o f caribou abundance is essential i f researchers are interested in obtaining further insight into the complex influences that climate imposes on this ungulate species. The two main objectives o f this study were to: (1) provide evidence for changing barren-ground caribou abundance during the last century, using dendroecological evidence, and (2) evaluate the correlation between large-scale climate patterns, such as the Arct ic Oscillation, on the cycles in population abundance o f these caribou. 73 4.2 METHODS 4.2.1 Study Area and Site Selection The study sites were situated along caribou migration trails near the treeline in the Northwest Territories between 61° N , 106° W and 65° N , 115° W (Fig. 2.1). The sites were located in the forest-tundra transition zone, with black spruce {Picea mariana [Mil l . ] BSP) being the dominant tree species (Matthews et al. 2001). Caribou tend to follow the same migration routes to and from their calving grounds, and their trails are imprinted on the landscape (Gunn et al. 2001). A total of nineteen sites were sampled for barren-ground caribou activity; sites 1-15 were selected based on information from Dogrib elders who identified areas across the treeline that were frequented by the Bathurst caribou herd (Dogrib Treaty 11 Council 2001). The remaining four sites (16-19) were selected on the late summer range o f the Beverly herd. Details of the sampling and analysis procedures were presented in Chapter 2 and in Zalatan et al. (2006). 4.2.2 Relationships between meteorological data and the AO I tested the relationship between the summer index of the A O (AOS) and the winter index of the A O ( A O W ) and the meteorological data (mean monthly temperature, total monthly precipitation and total monthly snow depth; Meteorological Service of Canada monthly climate data from Yellowknife airport, 62°27'N, 114°26'W, 205.70 m a.s.l.; http://climate.weatheroffice.ec.gc.ca/prods_servs/cdcd_iso_e.html) using Pearson's correlations. The data were truncated to the period between 1943-2001 to ensure that there were no missing data in the record. The purpose of this analysis was to determine i f the A O W and A O S have an impact on the local climate, and hence to better understand how the A O may influence caribou abundance cycles. 74 To evaluate the temporal stability of the shared variance among the monthly index o f the A O S , Principal Components Analyses were performed for successive overlapping 30-year periods (1900-1930, 1910-1940, 1920-1950, 1930-1960, 1940-1970, 1950-1980, 1960-1990, and 1970-2000) and for the entire common period (1900-2000) (Tardif et al. 2002). Since the standard period for studying climate trends is 30 years (30-year climate normals), I chose this same time period to evaluate the temporal stability of the A O . Analyzing the change in the variance of the A O may provide some insight to help further decipher the temporal dynamics between caribou and the A O . 4.2.3 Barren-ground caribou abundance estimates A total of 1991 trampling scars were found in Picea mariana root samples and included in the scar frequency distribution which represent a proxy-record for caribou abundance (Zalatan et al. 2006; Chapter 2). This reconstruction o f caribou abundance covers the period from 1900-2002, which is a much longer period that has been previously presented (Post and Stenseth 1999; Aanes et al. 2002; Griffith et al. 2002; Post and Forchhammer 2002; Hebblewhite 2005). These other studies obtained population data for caribou and reindeer for periods from 13 and 43 years. The shorter period of data for these caribou/reindeer populations is due to the fact that the data were derived from aerial photography or animal counts and can therefore be only as long as the data collection method permitted. Conversely, the benefit o f using dendroecology to reconstruct caribou abundance is that the record can be much longer. Most of the scars are formed during late summer/early fall, when the ground is snow-free. A s a consequence of the difficulty associated with dating these scars (Chapter 2), the data have been 75 grouped into 5-year age-classes, resulting in only 21 data points. There is an increasing underestimation of caribou abundance with increasing age of the roots. Various factors contribute to the loss of scars through time such as the death of scar-bearing roots, fading of scars by weathering, decomposition, and repeated caribou trampling activity (Morneau and Payette 1998, 2000). This underestimation was addressed by applying a log-linear regression and using the residuals to obtain a more accurate depiction o f the abundance pattern of these herds (Morneau and Payette 1998, 2000). Departures from the negative exponential model were used to demonstrate the years of high and low caribou abundance (Chapter 2; Zalatan et al. 2006). 4.2.3.1 Determining the relation between the AO and caribou abundance cycles I tested for the relationship between the A O S and the scar frequencies as a proxy for shifts in caribou abundance for the last two phases of the A O S separately (1955-1985 and 1990-2000) in 5-year age-classes, for these two periods combined (1955-2000) and for the entire study period (1900-2000) using Spearman's rank correlations. The dataset for the A O was grouped into five-year age-classes to perform equivalent correlations with the caribou residual data. Although the time series of the A O W did not follow any trends in caribou abundance, the last decade (1989-2002) of the A O W corresponded strongly to the scar frequencies (www.arctic.noaa.gov/detect/climate-ao.shtml). Thus, Spearman's rank correlations were also performed between the A O W and the caribou abundance cycles for the period between 1990 and 2000 (in 5-year age-classes). Correlations were also performed using annual data (1-year age-classes) between the caribou abundance cycles and the annual data of the A O W for the equivalent period (1989-2002) to determine whether higher resolution data would provide 76 different results. I used annual data for this analysis since the time period is too short to justify using a 5-year age-class, and therefore the results would be insignificant. 4.2.3.2 Cross wavelet transform and wavelet coherence analysis Another approach that has been used to examine the variability in the signal of time series data is wavelet analysis (Jevrejeva et al. 2003). Wavelet transforms are used to expand time series into time frequency space and are therefore useful for finding localized intermittent periodicities (Foufoula-Georgiou and Kumar 1995; Grinsted et al. 2004). The Continuous Wavelet Transform (CWT) is a common tool for analyzing oscillations in a time series, and is often used on two time series that are expected to be linked in some way. More specifically, the C W T examines whether there is a consistent phase relationship and, therefore, a suggestive causality between two time series. Wavelet transforms are often used to analyze time series that contain nonstationary power at many different frequencies (Jevrejeva et al. 2003). The Wavelet Coherence (WTC) is calculated from two C W T s , and represents the local correlation between two time series in time frequency space. Another way to define the W T C is that it helps determine how two time series are coherent even i f the common power (correlation) is low, and shows how confidence levels against random noise backgrounds are calculated. The advantage o f these methods is their ability to detect amplitude- and phase-modulated oscillations (Jevrejeva et al. 2003). Here I examined the connection between the A O S and caribou abundance cycles, specifically the phase relationships between the two time series using both C W T and W T C . The A O W was not examined in this analysis as it was not showing any relation to the caribou abundance cycles (see 77 Results). Wavelet transforms require as many data points as possible therefore I used annual caribou residual data rather than grouping the scars in 5-year age-classes (Chapter 2). 4.3 R E S U L T S 4.3.1 R e l a t i o n s h i p s b e t w e e n m e t e o r o l o g i c a l d a t a a n d t h e A O A positive correlation was found between the AOS and mean monthly summer temperatures, June-August (Fig. 4.1 A , r 2 = 0.43, n = 58, /K0.05) . The A O W was also positively correlated to summer temperatures, June-August (Fig. 4.IB, r 2 = 0.34, n = 58,/J<0.05). A . B . -60 -.10 40 Arctic Oscillation Index (Jun-Aug) -160 -110 -60 -10 40 90 140 Arctic Oscillation Index (Oct-May) F I G U R E 4 .1 (A) Relationship between the summer index of the standardized Arctic Oscillation (AOS) and mean monthly summer temperatures for the period 1943-2001; ( B ) Relationship between the winter index of the standardized Arctic Oscillation (AOW) and mean monthly summer temperatures for the period 1943-2001. 78 However, the AOS and A O W were statistically unrelated to both total monthly precipitation and total monthly snow depth. From this analysis, I concluded that the strongest relationship was found between the AOS/AOW and summer temperatures. As the highest correlation was found between the AOS and mean monthly temperatures for June to August, I proceeded by using the AOS only. The curves of the AOS and mean monthly summer temperatures show similar trends for the period 1943-2001, thus indicating the influence of the AOS on mean monthly summer temperatures in this region (Fig. 4.2) 17 < c 3 16 u 15 P. >-. 3 13 12 Mean Monthly Temperature (°C; Jun-Aug) Arc t ic Oscillation (Jun-Aug) -1.0 1950 1960 1970 1980 1990 Year 2000 F I G U R E 4.2 The standardized Arctic Oscillation (AO) anomaly index for June to August and mean monthly temperatures for June to August for the comparative period of 1943-2001. Pearson's correlations were used to analyze the relationship between only negative (<0) AOS index years and mean monthly temperature. This analysis was also performed using only positive (>0) AOS index years. The purpose of this analysis was to determine whether the local climate is 79 responding to both phases of the A O S (positive and negative) or only one of the phases of the A O S . Correlations computed over the entire period of analysis (1943-2001) with only positive and negative years of the A O S showed no significant correlations. However, negative values of the July monthly index of the A O were significantly correlated to July temperatures (r 2 = 0.43; n = 58; p O . 0 5 ) . The results from the P C A calculated for successive overlapping 30-yr periods indicated that the percentage of explained variance by the two first principal components was relatively constant until 1950 (Fig. 4.3). The explained variance of the first principal component remained between 18%-21% until 1980, however it increased to 24% between 1960-1990. The explained variance of the second principal component remained between 14%-16% until 1980, however it decreased to 13% between 1960-1990 These results demonstrated that the explained variance o f the A O did not fluctuate a great deal until approximately 1960, after which it became more variable. 4.3.2 Barren-ground caribou abundance cycles The scar frequency distribution from both groups of sites showed similar abundance cycles through time (Chapter 2; Fig . 4.4; r s = 0.55, n = 21, j?<0.05). Caribou abundance was high during the mid-1940s, and 1990s, and was very low during the 1920s, 1950s-70s and in the 2000 age-class. The results from the scar frequency distribution correlate strongly with trends observed in the data from aerial surveys and traditional knowledge (Dogrib Treaty 11 Council 2001; Zalatan et al 2006). 80 24 o o t 20 16 12 1900-1930 19.10-1940 1920-1950 1930- 1940-1960 1970 Time span 1950-1980 1960-1990 1970-2000 F I G U R E 4.3 Percent variance expressed by the first and second principal components of PCA based on a matrix with the monthly index of the Arctic Oscillation for the subperiods: 1900-1930, 1910-1940, 1920-1950, 1930-1960, 1940-1970, 1950-1980, 1960-1990, and 1970-2000. For the whole period 1900-2000, the variance explained by the first and second principal components is 50.3%. 1940 1960 1980 Year (Five-year age-classes) 2000 F I G U R E 4.4 Standardized distribution of the number of trampling scars (residuals of the log-linear regression) from 1900-2000 for the northwest and southeast sites. 81 4.3.3 Determining the relation between the AO and caribou abundance cycles There was no correlation found between the A O S and the standardized residuals o f the northwest and southeast sites for the entire study period between 1900 and 2000 (r s = -0.07, n = 21,/?>0.05; r s - -0.21, n = 21, p>§.§5, respectively). During the first positive phase of the A O S , caribou abundance was low (Fig. 4.5). A s the A O S began the switch to a negative phase (1931-1955), caribou abundance began to increase. When the A O S was in a slightly negative phase, caribou abundance was high. During the following positive phase o f the A O S (1956-1983), caribou abundance began to decrease once again, however not as much as during the first positive phase. During the final negative phase of the A O S (1984-2000), the caribou abundance became more variable with little or no trend at all in response to the A O S . 1.0 0.5 o •*= 0.0 -0.5 -1.0 Positive Phase (1900-1930) Caribou abundance at northwest sites Caribou abundance at southeast sites 5th order curve of the A O (Jun-Au'g) Negative Phase (1984-2000) 'Negative Phase! (1931-1955) Positive Phase (1956-1983) rt 3 N -5 nt TD C & F-2 1900 1920 1940 1960 1980 2000 Year F I G U R E 4.5 Residuals of the log-linear regression on the frequency distribution o f trampling scars for both sets of sites grouped into five-year age-classes and a 5 t h order regression curve o f the summer index of the Arctic Oscillation (June-August) for the period 1900-2002. 82 Although they were not significant, the positive correlations between the A O S and the caribou residuals for the last positive phase (1955-85) were high (northwest sites: r s = 0.71, n = 7, p > 0.05; southeast sites: r s = 0.54, n = l,p> 0.05). The correlations between the caribou residuals and the A O S for the last negative phase (1990-2000) were high, but were not significant (northwest sites: r s = 0.50, n = 3, /»>0.05; southeast sites: r s = -0.50, n = 3, p> 0.05). The correlations between the A O S and the caribou residuals for the period from 1955-2000 were relatively high but not significant (northwest sites: r s = 0.39, n = 10,/?>0.05; southeast sites: r s = 0.49, n = \Q,p> 0.05). The last positive and negative phases of the A O S did not have an inverse relation to the caribou abundance as was seen in the first two phases. Both the caribou abundance and the A O S were more variable and followed roughly similar trends from 1955-2000. The last decade (1989-2002) of the A O W time series was strongly positive (Fig. 4.6). 1900 1920 1940 1960 1980 2000 F I G U R E 4.6 The standardized Arctic Oscillation (AO) anomaly index for October to M a y for the period between 1900-2002, (data from http://tao.atmos.washington.edu/ao/#monthly). 83 Correlations were not significant between the A O W and caribou abundance (using 5-year age-classes) at either the northwest (r s = -0.5, n = 3; /?>0.05) or southeast sites (r s = 0.5, n - 3; p>0.05). Using annual data, correlations were also not significant between the A O W and northwest sites (r s = -0.12, n = 13; p>0.05) or the southeast sites (r s = 0.39, n = 13; p>0.05). 4.3.3.1 Wavelet Transform Analysis The Cross Wavelet Transform (CWT) between the two groups of caribou abundance cycles and the A O , as well as between the two groups o f caribou abundance are presented in Figures 4.7, 4.8, and 4.9. The thick black contour designates the 5% significance level against random noise and the cone of influence (COI) where edge effects might distort the picture is shown as a lighter shade. The y-axis is the frequency of the dominant cycle, the x-axis represents the years in which that cycle dominates and the color represents the strength o f the relationship. The C W T s are not a reliable indication o f causality, but rather depict similarities in periodicity through time. They are useful to illustrate patterns in time series curves in comparison to other time series. For all remaining wavelet transform figures, the area above the 5% significance level is not a reliable indication of causality and therefore the results outside of these areas should be interpreted with caution. The time-frequency patterns for the northwest caribou abundance and the A O S (Fig. 4.7) were not perfectly coherent, however, a 16-year oscillation was detected between around 1930 and 1970 in both time series. The northwest caribou abundance had a periodicity o f 2.5 years at 1930 and mid-1950s. The A O S had a 2.5-year periodicity in the 1930s, a 4-year periodicity in the early 1900s, 1970s and again 1990s, and an 8-year periodicity in the 1990s. These findings are 84 similar to those found in the winter index of the A O (2.2-2.4-, 7.8-, and 12.8-year periodicities; Jevrejeva et al. 2003). The C W T of the northwest caribou abundance and the A O S depicted some similarities in periodicity through time. nwr 1900 1920 1940 1960 1993 2000 F I G U R E 4.7 The cross wavelet transform (CWT) of the northwest caribou abundance (top panel) and the A O S (bottom panel). Similar to Fig. 4.7, there was a hint of a 16-year period between approximately 1930 and 1970 in the southeast caribou abundance wavelet transform (Fig. 4.8). This time series also experienced a 2.5-5- year periodicity between 1920 and 1935. Although this is within the cone of influence (COI), where edge effect might distort the results, the C W T of the two caribou populations demonstrated common features between the two time series o f 2.5-5-year periodicity around 2000 (Fig. 4.9). 85 ser 1900 1920 1 940 1 993 1 980 2000 A O ~ - I i i i _ J 1900 1 920 1 940 1 960 1 980 2000 F I G U R E 4.8 The cross wavelet transform (CWT) of the southeast caribou abundance (top panel) and the AOS (bottom panel). ser 1900 1920 1940 1960 1980 2000 F I G U R E 4.9 The cross wavelet transform (CWT) of the northwest caribou abundance (top panel) and the southeast caribou abundance (bottom panel). 86 The figures of the wavelet coherence (WCT) can be interpreted in a similar way as the C W T s , with directional vectors that depict in-phase (vectors pointing up or to the right) and anti-phase (vectors pointing down or to the left) relationships between two time series. In the W T C of both caribou abundance time series, several sections are significant and these areas are mostly in-phase (Fig. 4.10). There was a 2, 4-5, 2-5, and 14 year periodicity found between 1930 and 1980. The benefit of the W T C is that it finds not only high common power but also similar patterns in behavior of time series during specific time periods (or phases), and clearly the W T C of both caribou abundance cycles demonstrated the similarity between the two time series. This is not surprising based on the results from Chapter 2 which showed that the residuals of the two groups of sites (northwest and southeast) were positively correlated (r s = 0.55). WTC: ser-nwr i _ i ) v *• * *• *• * . * I ^ S I I i I 1900 1 920 1 940 1 980 1 993 2000 F I G U R E 4.10 The wavelet coherence (WTC) between the southeast caribou abundance and the northwest caribou abundance time series. 87 The W T C between southeast caribou abundance and northwest caribou abundance demonstrated 3-5 year periodicity during the 1930s, followed by a periodicity of between 2-5 year during the 1940s until 1950. There was a 15-16 year periodicity during the 1960s and following this there was strong coherence from 1980 until 2000 (although most of it is within the COI). The northwest caribou abundance and the A O showed high coherency in the 2-2.5- and 8-12-year bands, with rather low coherency between 1900-1930 and 1960-1980 (Fig. 4.11). WTC: nwr-AO J 1 I ^^^^r i I 1 1 1900 1 920 1 940 1 960 1 930 2000 F I G U R E 4.11 The wavelet coherence (WTC) between the northwest caribou abundance and the A O S time series. Although it is not significant (it is outside the 5% significance level (black contour line) and the power is low), there was once again a hint of a 16-year periodicity between 1930 and 1960. The vectors indicated an anti-phase relationship (vectors were pointing down and to the left) in the early 1900s, and from 1930-1960, after this period the relationship became in phase (vectors 88 were pointing up and to the right). The southeast caribou abundance and the A O S showed high coherency in the 2-2.5-, 8-12-, and 16-year bands, with low coherency in the 1920s, and between 1970-1990 (Fig. 4.12). W T C : s e r - A O 1900 1920 1940 1980 1980 2000 F I G U R E 4.12 The wavelet coherence (WTC) between the southeast caribou abundance and the A O S time series. The strong anti-phase periodicity in the 8-12-year band was consistent with the W T C of the northwest caribou abundance and the A O S . Other similarities can be seen such as the anti-phase 4-year band in the early 1900s, the in-phase 2-year band (although not significant) in the 1990s. The 16-year periodicity between 1930 and 1960 was much stronger in this W T C . Again, the periodicities were anti-phase in the early 1900s, and during the 1960s (8-12-year periodicity), however the most recent periodicity, between 1990 and 2000, (although it is in the COI) was in-phase. 89 4.4 DISCUSSION This study demonstrates important correlations between decadal climatic patterns and abundance of barren-ground caribou. Specifically, it illustrates the complexity associated with relating large-scale Arctic climate variation with abundance cycles of barren-ground caribou at multi-decadal time scales. Unlike other studies that have shown a positive relation between large-scale climate and an ungulate population (Post and Stenseth 1998, 1999; Post et al. 1999; Patterson and Power 2001; Aanes et al. 2002; Griffith et al. 2002; Hebblewhite 2005), my study demonstrates the variability and uncertainty associated with relating mammals to climate without having a dataset that covers multi-decadal time frames. These other studies have compared climate to ungulate population abundance within a much shorter time frame (between 13 and 43 years). M y research is the first to compare long-term (100 years) ungulate population data with a large-scale climate pattern. M y analyses showed that the relation between the A O S and caribou abundance has not remained constant. . The wavelet analysis provided a means to follow gradual changes in the natural frequency of the A O and caribou abundance time series. The wavelet coherence (WTC) demonstrated that from about 1900-1930 (first positive phase of the A O S ) , the A O S and caribou abundance cycles were in anti-phase. Similarly, from 1930-60 (negative phase of the A O S ) , the A O S and the caribou abundance cycles were also in anti-phase. Thus, for the first two phases there was an inverse relationship between the caribou abundance cycles and the A O S . The A O S and caribou abundance cycles changed to a significant in-phase relationship from approximately 1970-2000 (the most recent positive and negative phases). Thus, it was evident that a phase switch in the A O S may have been a factor in the increase or decrease in the abundance of these barren-ground 90 caribou over time. The W T C demonstrated that the relation between A O S and caribou abundance cycles is not linear and has not remained constant during the last 100 years. There was a strong relationship between the July index of the A O and tree-growth at most sites (Chapter 3), which has important indirect implications for caribou. This relationship can be explained by the influence of the A O S on local climate as the mean temperatures during July and August were correlated to the A O S for the period between 1943-2001. These relationships are not surprising since many studies have been able to model the relationship between large-scale climate and temperature (e.g. Thompson et al. 2000). The positive phase of the A O S results in decreased temperatures and increased precipitation during the summer months. It is possible that a positive A O S may be associated with unfavorable forage growth conditions (cooler and wet weather) during the growing season, which could help explain the low abundance of caribou during the first positive phase. Forage quality on caribou ranges has been shown to be influenced by climate (Chapin et al. 1995). Caribou are selective in their foraging habits during spring and summer, choosing plants that are high in nutrients and low in secondary compounds (Klein 1970; Bryant et al. 1983; White 1983). Changes in climate conditions can alter nutrient concentrations, anti-herbivore defenses, and forage availability (Chapin et al. 1995). Field experiments have simulated environmental changes (e.g. increased temperatures, reduced irradiance, reduced precipitation) to determine the effect on growth and nutrient content in several plant species that are important forage for caribou (Chapin and Shaver 1985; Shaver et al. 1986; Chapin et al. 1995). Forage quality and availability directly influence body condition of female caribou, affecting the unborn calf (Skogland 1985). Thus, poor nutrition from low quality and availability of forage can result in reduced numbers of caribou (Valkenburg et al. 1996). 91 Following the first positive phase, caribou abundance was high (during the negative phase of the A O S ; 1931-1955), when winters tend to be cold and dry. These conditions are conducive to reduced snow pack, which has been linked to reduced rates o f predation (Nelson and Mech 1986; Huggard 1993; Post et al. 1999; Hebblewhite et al. 2002; Hebblewhite 2005). Reduced snow pack can also provide easier access to forage, which is l ikely more important (Turner et al. 1994). During the following two phases of the A O S (1956-1983 and 1984-2000), the time series of both the caribou abundance and the A O S were no longer out of phase but were more coherent. The inverse synchrony that was seen between 1900 and 1955 was no longer present. The caribou abundance was affected by this change since the abundance series also lacked any major fluctuations, with the exception of the dramatic drop at the turn of the 21 s t century. The trends in the winter index of the A O demonstrated highly positive values during the last phase, however correlations between the A O W and caribou abundance were not significant during this period. Thus, the A O W was not l ikely having an important impact on caribou abundance at the turn of this century. There are other factors which could help explain changes in caribou abundance. In areas without a major influence of predation, density-dependence and environmental stochasticity become major determinants for ungulate population dynamics (Saether 1997; Gaillard et al. 1998). High animal densities and adverse weather have been shown to influence the population numbers of ungulates (Clutton-Brock et al. 1987; Singer et al. 1997). Increased mosquito abundance (Downes et al. 1986; Noel et al. 1998) and a change in the pattern of winter predation (Nelson 92 and Mech 1986; Huggard 1993; Post et al. 1999) have also been linked to decreases in caribou numbers. Further research would be needed to determine the interaction among these factors, which is beyond the scope of this research. The P C A of overlapping periods on the monthly A O index demonstrated that the A O underwent a distinctive change in the 1950s, during the most recent positive and negative phases. A recent study has shown that the A O has had a different, more episodic behavior, including multi-year runs of positive or negative anomalies, during the last decade (1996-2004) with values being more negative or neutral than positive (Overland and Wang 2005). These trends correspond to the most recent phase of the A O S which was negative (1984-2000). Stratospheric temperature anomalies, another index of the strength o f the polar vortex, have also shown this episodic character over the Arctic (Rind et al. 2005). This topic remains a central research question in the field of synoptic climatology and renders predictions about the direction o f the A O relatively uncertain. One could put forth the argument that the inconsistent relation between climate and these caribou abundance patterns over longer time scales indicates that caribou are not responding to the A O in this region. However, I would reject this suggestion because the data show that the first two phase-shifts of the Arctic Oscillation are negatively, although weakly, correlated to increases or decreases in caribou abundance. This indicates that there is a large-scale process influencing these herds but that this relation appears to have changed during the most recent positive and negative phases (1955-1983 and 1984-2000). This simply means that the relationship has changed through time rather than there being no relationship at all . This illustrates the most 93 important aspect of this study, which is that the relationship between the A O and caribou abundance has not remained consistent when longer time periods are analyzed. Although there was not a constant relationship through time, the A O must be driving changes in the abundance cycles o f these barren-ground caribou because of its important influence on weather patterns and climate in the Arctic. The difficulty in this type of study is that there is a trade-off between higher resolution short-term data and lower resolution long-term data. Without the long-term data generated in this study, I could not conclude analyses at multiple time scales and we would be making incorrect interpretations about the relation between the A O and abundance o f barren-ground caribou. 4.4.1 Predicting the impacts of climate change This study is the first to demonstrate the long-term relation between the A O and cycles in barren-ground caribou abundance. This relationship is not linear over long time scales and may therefore be very difficult to predict in the future. Mysterud et al. (2001) also noted the difficulty in predicting the ecological impacts of large-scale climate fluctuations, and emphasized the need to consider non-linear relationships. Recent studies are now quantifying nonlinear ecological effects of climate variability (e.g. Stenseth et al. 2002). Model projections of Arctic temperatures have suggested large region-to-region variability in the future response o f atmospheric circulation to external forcing (Overland and Wang 2005). The use of teleconnections as a forecast tool is only useful i f it provides insight into the interaction between prevailing meteorological conditions in a consistent manner over time (Creilson et al. 2005). If the phases of the A O remain stable and prominent, the impact on the 94 local meteorological conditions wi l l be more predictable. Thus, the interaction between climate and barren-ground caribou could simultaneously be more predictable. However, a recent study has noted an increase in the positive phase of the North Atlantic Oscillation ( N A O ) during the second half of the 20 t h century, which has resulted in a greater number of winter warm spells over most of Canada (Shabbar and Bonsai 2003). Wi th the onset of global warming, the A O has already demonstrated more variable and episodic behavior ( A C I A 2004), thus rendering predictions about the future impact on caribou populations even more difficult. The change in the response of caribou to climate during the most recent phases of the A O may also be compounded by other factors such as predation and hunting. In addition, the recent change in the A O , which had been linked to global warming, may result in caribou responding differently to changes in climate during a decline in abundance compared to an increase in abundance. Thus, these factors need to be addressed through further research. The experience of the last decade suggests that researchers should exercise considerable caution in addressing Arctic change and the response of caribou to climate. 95 4.5 R E F E R E N C E S Aanes, R., Sasther, B . , Smith, F . M , Cooper, E.J . , Wookey, P .A . , and 0ritsland, N . A . 2002. The Arct ic Oscillation predicts effects of climate change in two trophic levels in a high-arctic ecosystem. Ecology Letters, 5: 445-453. Arctic Climate Impact Assessment. 2004. Arctic Climate Impact Assessment. Cambridge: Cambridge University Press. Boudreau, S., Payette, S., Morneau, C , and Couturier, S. 2003. Recent decline of the George River caribou herd as revealed by tree-ring analysis. Arctic, Antarctic, and Alpine Research, 35: 187-195. Bryant, J.P., Chapin, F.S., III, and Kle in , D.R. 1983. Carbon/nutrient balance of boreal plants in relation to vertebrate herbivory. Oikos, 40: 357-368. Caughley, G . , and Gunn, A . 1993. Dynamics of large herbivores in deserts: kangaroos and caribou. Oikos, 67: 47-55. Chapin, F.S., III, Shaver, G.R. , Gibl in , A . E . , Nadelhoffer, K . , and Laundre, J .A. 1995. Responses of arctic tundra to experimental and observed changes in climate. Ecology, 76: 694-711. Chapin, F.S., III, and Shaver, G.R. 1985. Individualistic growth response of tundra plant species to environmental manipulations in the field. Ecology, 66: 564-576. Clutton-Brock, T . H . , Major, M . , Albon, S.D., and Guiness, F . 1987. Early development and population dynamics in red deer. I. Density-dependent effects on juvenile survival. Journal of Animal Ecology, 56: 53-67. Creilson, J .K. , Fishman, J., Wozniak, A . E . 2005. Arctic Oscillation-induced variability in satellite-derived tropospheric ozone. Geophysical Research Letters, 32: 1-5. 96 Dogrib Treaty 11 Council . 2001. Caribou migration and the state of their habitat. Final Report to the West Kitikmeot/Slave Study Society, Yellowknife, N T . Downes, C M . , Theberge, J .B. , and Smith, S . M . 1986. The influence of insects on the distribution, microhabitat choice, and behavior of the Burwash caribou herd. Canadian Journal of Zoology, 64: 622-629. Environment Canada. 2005. Climate trends and variations bulletin. Temperature and precipitation in historical perspective. Annual 2003. Climate trends and precipitation bulletin [online]. Available from http://www.msc-smc.ec.gc.ca/ccim/bulletiriyarmual03/regional_e.cfrn [accessed October 31,2005] Forchhammer, M . C . , Post, E . , Stenseth, N . C . , and Boertmann, D . M . 2002. Long-term responses in arctic ungulate dynamics to changes in climatic and trophic processes. Population Ecology, 44: 113-120. Foufoula-Georgiou, E . , and Kumar, P. (eds.) 1995. Wavelets in Geophysics. San Diego: Academic. Gaillard, J . - M . , Festa-Bianchet, M . , and Yoccoz , N . G . 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution, 13: 58-63. Griffith, B . , Douglas, D .C . , Walsh, N . E . , Young, D .D . , McCabe, T.R., Russell, D .E . , White, R . G . , Cameron, R .D . , and Whitten, K . R . 2002. Arctic Refuge Coastal Plain Terrestrial Wildlife Research Summaries: Section 3: The Porcupine Caribou Herd - Part J. U . S . Geological Survey Biological Science Report: USGS/BRD/BSR-2002-0001. Alaska: U .S . Geological Survey. 97 Grinsted, A . , Moore, J.C., and Jevrejeva, S. 2004. Application o f the cross wavelet transform and wavelet coherence to geophysical time series. Nonlinear Processes in Geophysics, 11: 561-566. Gunn, A . 2005. Voles, lemmings and caribou - population cycles revisited? Rangifer Spec Issue, 14:105-112. Gunn, A . , Dragon, J., and Boulanger, J . 2001. Seasonal movements of satellite-collared caribou from the Bathurst herd. Final Report to the West Kitikmeot Slave Study Society. West Kitikmeot Slave Study Society: Yellowknife, N W T . http://www.wkss.nt.ca/HTML/08_ProiectsReports/PDF/SeasonalMovementsFinal.pdf Hebblewhite, M . 2005. Predation by wolves interacts with the North Pacific Oscillation (NPO) on a western North American elk population. Journal of Animal Ecology, 74: 226-233. Hebblewhite, M . , Pletscher, D . H . , Paquet, P .C . 2002. E lk population dynamics in areas with and without predation by recolonizing wolves in Banff National Park, Alberta. Canadian Journal of Zoology, 80: 789-799. Huggard, D.J . 1993. Effect of snow depth on predation and scavenging by gray wolves. Journal of Wildlife Management, 57: 382-388. Jevrejeva, S., Moore, J.C., and Grinsted, A . 2003. Influence of the Arctic Oscillation and E l Nino-Southern Oscillation (ENSO) on ice conditions in the Baltic Sea: The wavelet approach. Journal of Geophysical Research, 108, No . D21, 4677, doi: 10.1029/2003JD003417. Kle in , D .R. 1970. Tundra ranges o f the boreal forest. Journal of Range Management, 23: 8-14. 98 Matthews, S., Epp, H . , and Smith, G . 2001. Vegetation classification for the West Kitikmeot/Slave Study region. Final report to the West Kitikmeot/Slave Study Society. West Kitikmeot/Slave Study Society: Yellowknife, N W T . Morneau, C , and Payette, S. 1998. A dendroecological method to evaluate past caribou (Rangifer tarandus L.) activity. Ecoscience, 5: 64-76. Morneau, C. , and Payette, S. 2000. Long-term fluctuations of a caribou population revealed by tree-ring data. Canadian Journal of Zoology, 78: 1784-1790. Mysterud, A . , Stenseth, N . C . , Yoccoz, N . G . , Langvatn, R., and Steinheim, G . 2001. Nonlinear effects o f large-scale climatic variability on wi ld and domestic herbivores. Nature, 410: 1096-1099. Nelson, M . E . , and Mech, L . D . 1986. Relationship between snow depth and gray w o l f predation on white-tailed deer. Journal of Wildlife Management, 50: 471-474. Noel , L . E . , Pollard, R . H . , Ballard, W . B . , and Cronin, M . A . 1998. Act ivi ty and use of active gravel pads and tundra by caribou, Rangifer tarandus grand, within the Prudhoe Bay O i l Field, Alaska. Canadian Field-Naturalist, 112: 400-409. Overland, J.E., and Wang, M . 2005. The Arctic climate paradox: The recent decrease of the Arctic Oscillation. Geophysical Research Letters, 32: 1-5. Overpeck, J., Hughen, K . , Hardy, D . , Bradley, R., Case, R., Douglas, M . , Finney, B . , Gajewski, K . , Jacoby, G . , Jennings, A . , Lamoureux, S., Lasca, A . , MacDonald, G . , Moore, J., Retelle, M . , Smith, S., Wolfe, A . , and Zielinski, G . 1997. Arctic environmental change of the last four centuries. Science, 278: 1251-1256. 99 Patterson, B .R . , and Power, V . A . 2001. Contributions of forage competition, harvest, and climate fluctuation to changes in population growth of northern white-tailed deer. Oecologia, 128: 244-255. Portier, C , Festa-Bianchet, M . , Gaillard, J . - M . , Jorgenson, J.T., and Yoccoz , N . G . 1998. Effects o f density and weather on survival of bighorn sheep lambs (Ovis canadensis). Journal of Zoology, 245: 271-278. Post, E . , and Forchhammer, M . C . 2002. Synchronization o f animal population dynamics by large-scale climate. Nature, 42: 168-171. Post, E . , and Stenseth, N . C . 1999. Climatic variability, plant phenology, and northern ungulates. Ecology, 80: 1322-1339. Post, E . , and Stenseth, N . C . 1998. Large scale climatic fluctuations and population dynamics o f moose and white-tailed deer. Journal of Animal Ecology, 67: 537-543. Post, E . , Peterson, R.O. , Stenseth, N . C , and McLaren, B . E . 1999. Ecosystem consequences of w o l f behavioral response to climate. Nature, 401: 905-907. Rind, D . , Perlwitz, J. and Lonergan, P. 2005. A O Y N A O response to climate change: 1. Respective influences of stratospheric and tropospheric climate changes. Journal of Geophysical Research, 110: D12107, doi:10.1029/2004JD005103. Saether, B - E . 1997. Environmental stochasticity and population dynamics o f large herbivores: a search for mechanisms. Trends in Evolutionary Ecology, 12:143-149. Shabbar, A . , and Bonsai, B . 2003. A n assessment of changes in winter cold and warm spells over Canada. Natural Hazards, 29:173-188. 100 Shaver, G.R. , Chapin, F.S., III, and Gartner, B . 1986. Factors limiting seasonal growth and peak biomass accumulation in Eriophorum vaginatum in Alaskan tussock tundra. Journal of Ecology, 74: 257-278. Singer, F.J . , Halting, A . , Symonds, K . K . , and Coughenour, M . B . 1997. Density dependence, compensation and environmental effects on elk calf mortality in Yellowstone National Park. Journal of Wildlife Management, 61:12-25. Skogland, T. 1985. The effects of density-dependent resource limitation on the demography o f wi ld reindeer. Journal of Animal Ecology, 54: 359-374. Stenseth, N . C , Mysterud, A . , Ottersen, G . , Hurrell, J .W., Chan, K . - S . , and Lima, M . 2002. Ecological effects of climate fluctuations. Science, 297: 1292-1296. Tardif, J., Conciatori, F., and Bergeron, Y . 2002. Comparative analysis of the climatic response of seven boreal tree species from northwestern Quebec, Canada. Tree-Ring Research, 57: 25-37. Telfer, E.S. , and Kelsal l , J.P. 1984. Adaptation of some large North American Mammals for survival in snow. Ecology, 65: 1828-1834. Thompson, D .W.J . , and Wallace, J . M . 2001. Regional climate impacts of the Northern Hemisphere annular mode. Science, 293: 85-89. Thompson, D.W.J . , Wallace, J . M . , and Hegerl, G . C . 2000. Annular modes in the extratropical circulation. Part II: trends. Journal of Climate, 13: 1018-1036. Turner, M . G . , W u , Y . , Wallace, L . L . , Romme, W . H . , and Brenkert, A . 1994. Simulating winter interactions among ungulates, vegetation, and fire in northern Yellowstone Park. Ecological Applications, 4: 472^196. 101 Valkenburg, P., Davis, J .L. , Ver Hoef, J . M . , Boertje, R . D . , McNay , M . E . , Eagan, R . M . , Reed, D.J . , Gardner, C . L . , and Tobey, R . W . 1996. Population decline in the Delta caribou herd with reference to other Alaskan herds. Rangifer, Special Issue N o . 9: 53-62. White, R . G . 1983. Foraging patterns and their multiplier effects on productivity o f northern ungulates. Oikos, 40: 377-384. Zalatan, R., Gunn, A . , Henry, G .H.R . 2006. Long-term abundance patterns of barren-ground caribou using dendroecology. Arctic, Antarctic and Alpine Research, In Press, September 2006. 102 CHAPTER 5. THESIS SUMMARY AND SYNTHESIS The main focus of this dissertation was to determine the relation between long-term caribou abundance cycles and climate in the forest-tundra of central Northwest Territories. Caribou are a vital part o f Arctic terrestrial ecosystems, yet little is known of their long-term abundance cycles. Using dendroecology, caribou abundance cycles of two groups o f barren-ground caribou were reconstructed from 1900 to 2000. There were synchronous trends in the cycles of these two groups of herds, which strongly correlated to data obtained from traditional knowledge of Dogrib elders in the region and animal counts based on aerial photography. The strength of this reconstruction provided robust information for investigating the spatio-temporal patterns in caribou abundance cycles throughout the range o f the barren-ground herds. Once the long-term abundance cycles of caribou were reconstructed, the next objective was to determine whether climate was influencing these cycles. However, the only annually-resolved climate data in the region is the short record available from Yellowknife. To increase the knowledge of climate variability in this region, I developed a series of tree-ring chronologies and a reconstruction of summer temperatures at seven of the 19 study sites where I had reconstructed caribou abundance. The radial growth of trees at the seven sites in this region demonstrated a common climate-growth signal. The July-August temperature reconstruction (1831-2002) explained 40% of the climatic variance in the instrumental data and was positively correlated to the meteorological records of July-August temperatures. The reconstruction demonstrated below-average temperatures until approximately 1900, followed by a slow increase in temperature until the early 1930s, after which summer temperatures became increasingly variable until the turn of 103 this century. L o w temperatures at the end of the Little Ice Age were seen in both the tree-ring chronologies and the July-August reconstruction. This reconstruction is an invaluable source of climate information since it is the first dendroclimatic study developed in this region of central N . W . T . Knowledge o f the response of tree-growth to climate at these sites provided a more local-scale perspective of climate variability in this region. I then explored the impact of large-scale climate, Arctic Oscillation (AO), on caribou abundance cycles. Most studies that aim to explain the relation between large-scale climate oscillations and mammal populations are limited in the length of their data because it is often derived from aerial surveys, which in most cases do not predate the 1970s. Thus, the results of most of these studies are that large-scale climate oscillations are linearly correlated to mammal population abundance cycles. M y study has demonstrated that a long-term perspective of the relation between caribou abundance cycles and climate is quite different than what can be obtained from short-term records. July-August temperatures had the greatest consistent effect on tree-ring growth at all sites. Similarly, the A O S was most highly correlated to summer temperatures (June-August). Thus, the A O S is having the greatest impact on summer temperatures, and the summer temperatures are driving tree-ring growth in this region. Caribou abundance cycles were closely, but inversely, related to the first two phases of the A O S (1900-1930 and 1931-1955). In contrast, caribou abundance was not inversely related to the most recent positive and negatives phases of the A O S (1955-1983 and 1984-2000), but rather weakly followed trends in the A O S . The first and second principal components derived from the P C A of the monthly A O index confirmed that there was a 104 change in the A O record during the last two phases. This was also seen through the changed response of the caribou abundance cycle during the most recent phases. Since the summer index of the A O seems to be most influential to the vegetation and to the local climate, it is most probable that any changes in this summer index wi l l affect the vegetation, which wi l l directly and indirectly affect caribou populations. Changes in the A O S could potentially affect the interaction between climate and fire frequency, which could potentially change the migration routes of caribou in autumn, thus exposing them to increased predation. Understanding the impacts o f such a change in climate dynamics is essential i f caribou populations in the N . W . T are to be properly managed. This dissertation was unique in that it demonstrated the nonlinear relationship between long-term caribou abundance cycles and large-scale climate. Without a long-term record of caribou abundance, I would not have been able to illustrate these variable, non-linear trends that were clearly illustrated in the data. Therefore, researchers should exercise caution when studying short-term relations between large-scale climate and mammal populations. 105 A P P E N D I X A The residuals of the log-linear regression for both groups of sites were plotted using different age-classes (Fig. A . l to A.4) . The trends in the residuals at the different age-classes all demonstrate low caribou abundance in the 1920s, followed by a period of higher abundance in the 1940-60s. A t the annual age-class there is little change in the abundance cycles of caribou during the 1960s to the 1990s, whereas as the age-class is increased, there is more o f a trend towards low caribou abundance during this period. The drop in caribou abundance in the 2000 time period is seen i n the residuals at all age-classes. The residuals from both groups o f sites in all age-classes are positively correlated to each other, however the highest correlation was found using the largest age-class (Table 5). A t the 5-year age-class, the correlation between the residuals at both groups of sites is the greatest (Chapter 2; r s = 0.55; /?<0.05). T A B L E A . l Spearman's rank correlations between scar frequency distribution for both groups of sites (northwest and southeast) annually, in 2-year age-classes, 3-year age-classes and 4-year age-classes. Correlations are significant at p<0.05. Annual data 2-yr age-class 3-yr age-class 4-yr age-class r s 0.33 0.37 0.40 0.47 106 'Northwest sites -1.0 -1.5 1900 1920 ,1940 I960 1980 2000 Year of sear formation (1 -year age-classes) F I G U R E A . l Residuals of the log-linear regression on the scar-frequency distribution o f trampling scars for both groups o f sites in 1-year age-classes. 0.5 I -.- Northwest sites • Southeast sites 0.0 c3 w———* i ^ w f • i '55 -0.5 -1,0 -1,5--2.0 - H : r > 1 [ : • 1— 1 1900 1920 1940 1960 1980 2000 Year of scar formation (2-year age-classes) F I G U R E A . 2 Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups of sites in 2-year age-classes. 107 0,8 03 03 -0.7 — Northwest sites • -»-,--, Southeast:sites \ \ VV, • -.» "FT • -i; 1900 1920 1940 1960 1980 2000 Year of sear .formation (3-year age-classes) F I G U R E A.3 Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups of sites in 3-year age-classes. 1900 1920 1940 .1960 1980 • 2000 Year of scar formation (4-year age-classes) F I G U R E A.4 Residuals of the log-linear regression on the scar-frequency distribution of trampling scars for both groups of sites in 4-year age-classes. 108 T A B L E A.2 Pearson's correlations among the seven tree-ring chronologies and mean monthly temperature. A l l marked correlations are significant atp<0.05; n-59. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Annual Jan Feb 0.12 Mar 0.16 0.13 Apr 0.10 0.25 0.13 May 0.08 0.21 0.10 0.41 Jun 0.02 0.24 -0.08 0.23 0.41 Jul -0.03 -0.02 -0.19 0.03 0.26 0.44 Aug 0.07 0.07 -0.02 0.02 0.12 0.17 0.30 Sep 0.07 0.16 -0.03 0.14 0.16 0.01 0.16 0.17 Oct 0.13 0.17 -0.01 0.15 0.22 0.05 0.01 0.11 0.20 Nov 0.26 0.14 0.08 0.13 -0.01 -0.11 0.09 0.16 0.05 0.29 Dec 0.04 0.05 -0.02 0.05 -0.03 -0.18 -0.07 -0.16 0.17 0.11 0.30 Annual 0.51 0.56 0.34 0.54 0.48 0.26 0.18 0.23 0.34 0.45 0.57 0.37 site 6 0.10 0.00 -0.12 -0.08 0.13 -0.04 0.19 0.18 -0.09 -0.13 0.01 0.13 0.04 site 1 0.00 0.10 -0.24 -0.07 0.03 -0.03 0.24 0.25 -0.03 0.09 0.08 0.12 0.07 site 8 -0.07 0.11 -0.28 0.03 0.11 0.10 0.36 0.31 -0.10 0.10 -0.04 0.00 0.04 site 11 •0.28 0.21 -0.15 -0.09 0.07 0.11 0.34 0.21 -0.09 0.06 -0.05 0.03 -0.01 site 13 -0.26 0.22 -0.12 -0.17 0.07 0.00 0.26 0.30 0.00 0.08 -0.12 -0.03 -0.04 site 16 0.00 0.07 -0.11 0.03 0.14 0.04 0.31 0.33 -0.01 0.16 0.13 0.09 0.17 site 19 0.05 0.09 -0.18 -0.07 0.06 -0.05 0.19 0.22 -0.05 -0.05 0.07 -0.03 0.02 T A B L E A .3 Pearson's correlations among the seven tree-ring chronologies and total monthly precipitation. A l l marked correlations are significant atp<0.05; n = 59. PJan PFeb PMar PApr PMay PJun PJul PAug PSep POct PNov PDec PAnnual PJan PFeb 0.13 PMar 0.07 -0.02 PApr 0.02 0.11 0.22 PMay 0.09 0.05 -0.02 -0.13 PJun 0.02 0.08 -0.15 0.04 -0.06 PJul 0.03 0.09 0.16 0.04 0.01 0.14 PAug 0.01 -0.15 -0.01 0.09 0.00 -0.12 -0.09 PSep -0.08 0.12 0.10 0.06 0.19 -0.03 0.02 -0.07 POct 0.06 0.07 0.28 0.11 -0.07 0.02 -0.07 0.01 -0.11 PNov 0.19 0.01 0.08 -0.08 0.00 0.10 0.23 -0.09 0.08 0.22 PDec -0.02 0.05 0.34 0.03 0.14 -0.21 0.26 0.20 0.18 0.13 -0.06 PAnnual 0.24 0.19 0.36 0.25 0.26 0.25 0.49 0.45 0.29 0.34 0.33 0.46 site 6 -0.33 -0.02 -0.14 -0.10 0.04 0.32 0.14 0.18 0.00 -0.16 -0.12 0.04 0.11 site 1 -0.35 0.00 -0.27 -0.22 -0.03 0.22 0.12 0.14 0.00 -0.22 -0.09 -0.14 -0.04 site 8 -0.18 0.00 -0.23 -0.14 -0.01 0.13 0.01 0.24 -0.01 •0.29 -0.11 -0.07 -0.02 site 11 -0.33 -0.05 -0.27 -0.17 -0.17 0.16 -0.13 0.08 -0.04 -0.16 -0.01 -0.26 -0.21 site 13 •0.30 -0.02 -0.27 -0.17 0.05 0.14 -0.03 0.03 -0.01 -0.22 0.01 -0.14 -0.13 site 16 -0.18 -0.11 -0.22 -0.06 -0.08 0.01 -0.16 0.05 0.11 -0.25 -0.15 -0.01 -0.20 site 19 -0.11 0.02 -0.13 -0.07 -0.19 0.19 0.02 0.00 0.09 -0.09 -0.08 -0.08 -0.05 T A B L E A.4 Response function analysis between the monthly index of the A O and the seven tree-ring chronologies over the entire study period from M a y o f the previous growth year to August o f the current growth year. Marked correlations are significant at p <0.05. Site 1 Site 6 Site 8 Site 11 Site 13 Site 16 Site 19 Pmay 0.11 0.14 0.04 . -0.14 0.01 -0.04 0.09 Pjun -0.02 -0.04 -0.06 0.05 -0.03 -0.06 -0.05 Pjul 0.02 -0.04 0.00 0.12 0.03 0.01 0.05 Paug -0.02 0.06 0.01 -0.13 0.01 0.13 0.02 Psep 0.01 0.03 0.03 0.01 -0.04 0.04 0.07 Poet 0.00 0.08 0.09 -0.07 -0.04 0.03 0.04 Pnov -0.27 -0.21 -0.24 -0.17 -0.18 -0.02 -0.19 Pdec 0.07 0.05 0.05 0.06 0.04 0.04 0.05 Jan 0.08 0.13 0.12 0.04 0.01 -0.05 0.11 Feb -0.11 -0.18 -0.09 -0.12 -0.13 -0.11 0.00 Mar 0.00 -0.03 -0.01 0.02 0.06 0.00 0.04 Apr -0.11 -0.15 -0.13 -0.05 -0.13 -0.08 -0.15 May -0.01 0.05 0.10 0.09 0.02 0.22 0.13 Jun 0.14 0.12 0.02 0.17 0.11 -0.02. -0.07 Jul 0.18 0.15 0.22 0.26 0.21 0.28 0.21 Aug -0.07 -0.15 -0.09 -0.08 -0.13 -0.09 -0.17 T A B L E A . 5 Spearman's rank correlation between the July index of the A O and the seven chronologies for each phase. Marked correlations are significant at p <0.05. site 6 site 1 site 8 site 11 site 13 site 16 site 19 1900-1930 -0.06 -0.04 0.04 -0.12 -0.10 -0.17 -0.12 1931-1955 -0.09 0.02 -0.27 -0.32 -0.21 -0.10 0.14 1956-1983 -0.24 -0.34 -0.40 -0.48 -0.19 -0.44 -0.25 1984-2000 -0.07 -0.36 -0.20 -0.35 -0.45 -0.37 -0.31 111 

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.831.1-0092761/manifest

Comment

Related Items