Mesure et modélisation du taux d'effondrement et de dépérissement de la canopée du peuplier faux-tremble, en Alberta
|Advisor:||Cumming, Steven G.; Pothier, David|
|Abstract:||Dieback and collapse of mature stands dominated by trembling aspen (Populus tremuloides Michx.) have often been observed in North America. Recent studies indicate that drought, insect defoliation, activity of fungal pathogens and climate change are the major factors responsible for these phenomena in Western Canada. They are also observed in Quebec, where we find old aspen stands. The objective of this study is to determine the rates of aspen dieback and collapse in north-eastern Alberta and to determine their variability across the landscape in order to develop predictive statistical models based on stand attributes and climate moisture index. Forest inventory data from approximately 1978 and 1990 were used to produce a large sample of 6837 points (a statistical unit meaning a stand or a part of a stand). A subset of this sample where only pure or nearly pure aspen stands 60 years old or greater and whose height in the first inventory was greater than 6-12m were selected for analysis. The stand attributes height, age, density and the presence or absence of regeneration were compared between the two inventories to find possible successional pathways and identify the phenomena of concern. These stands attributes, along with a climate moisture index, forest cover type and the number of years between inventories, were used as covariates in logistic regression models to predict the aspen collapse and dieback. Our findings confirm the occurrence of aspen dieback and collapse: in a final sample of 1756 points, 171 are classified as having undergone dieback and collapse after 17 years, or approximately 9.73% of our final sample. The annual rate of joint collapse and dieback is 0.59%, while for dieback only it is 0.15%. Age, presence or absence of regeneration and the interaction of climate moisture index and age increase the probability of collapse and dieback. Stand density and climate negatively influence the two phenomena. A landscape level random effect with a variance of 0.83 (showing the importance of random effect influence) versus a residual variance of 0.92 was observed in the mixed models analysis. Stand age and density remain the best predictors of these two phenomena.|
|Document Type:||Mémoire de maîtrise|
|Open Access Date:||18 April 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.