Évaluation des coûts de déplacement de la martre d'Amérique dans une sapinière boréale aménagée
|Abstract:||In contrast to the ideal free distribution (Fretwell and Lucas 1969) framework, animals have to face numerous constraints when moving through landscapes. However, these constraints have rarely been integrated, let alone quantified, into habitat selection studies. Spatially-explicit models are increasingly used to make predictions about distribution of animal populations, but they are based on subjective estimations of movement parameters such as resistance to movement and perceptual range. Finally, most ecologists studying movements collect data by radio-telemetry, but this is an expensive method to which alternatives exist. The objectives of this study were to (1) develop an inexpensive approach to analyze movements from spatial distribution data, and (2) estimate value of two parameters (movement resistance and perceptual range) integrated in simulation models of American martens (Martes americana). Every winter from 2004 to 2008, marten tracks were located by snow-tracking. For each year, I designed 150 movement resistance scenarios differing by three spatial resolutions (5 m, 25 m, and 100 m), 10 resistance factors in open areas (range 4-1000), and five scenarios differing by edge effect and relative movement cost in young stands (age 11-30 y). With a Geographical Information System (GIS), ‘least-cost’ paths were generated across the study area, according to each cost scenario. Nearest-neighbor distances between points where simulated paths met transects and real marten tracks were used to assess simulation performance. In 78 % of cases (45 % significant), resistance scenarios performed better than corresponding null models (equal resistance in all habitats), suggesting that mosaic structure of our study area did limit movements of American martens. Adding intermediate movement values for young forests increased simulation performance to 93 % (53 % significant), suggesting that movement resistance in young stands was higher than in mature forest. All simulations with a pixel size of 100 m performed significantly better than null models, suggesting that marten perceptual range was at least 100 m. However, model performance was similar among hypothetical movement costs in the open (within scenarios), preventing a reliable estimate of relative movement cost in the open. Studying movements with snow-tracking data appears promising, and may yield useful results in highly fragmented landscapes.|
|Document Type:||Mémoire de maîtrise|
|Open Access Date:||16 April 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.