13 時すぎから,
「動物の場所えらび (GPS 欠測も考慮)」 (habitat selection)
の統計モデリングについてのセミナー
(その 論文).
内容は下記のごとしで.
日時: 2013 年 9 月 4 日 (水) 13:15 から 1 時間程度
会場: 北海道大学・環境科学院 (http://www.ees.hokudai.ac.jp) D101 教室
講演者: Bryan Manly and Lyman McDonald (Western EcoSystems Technology, Inc.)
演題: Estimating Habitat Selection When the GPS Fix Success Rate is less than 100%
要旨:
This presentation is about situations where the plan is to detect a
radio-tagged animal at fixed intervals of time (e.g., every 6 hours)
so that its location is known in the area occupied and, based on this
information, estimate how the animal selects habitat based on the
available habitat and the detected locations. For example, if the area
of interest is divided up into square quadrats then given the quadrat
location of the animal for one detection it is possible to estimate a
function that gives the probability that the animal will be seen next
at any of the other quadrats in the area based on the habitats in the
quadrats and the distance from the first location. This would be
relatively easy to model if the animal is always detected when it is
supposed to be. If this is not the case then the probability of
detection is likely to depend on the habitat. We therefore propose a
method for estimating a resource selection function that says how the
animal prefers different habitats, and also estimating a detection
function that says how the probability of detection depends on the
habitat as well.