To return to the Midwest Fish & Wildlife Conference website, go to http://www.midwestfw.org/ The following schedule and room names are subject to change (as of February 1, 2017). Please check back for updates.
Presenters: Presenters for technical presentations are either the primary author (the first name listed in the abstract), or are indicated with an asterisk next to their name. Please note: the conference schedule is hosted by Sched.org which allows you to search within the schedule, and filter the schedule to show sessions only occurring on a certain date or within a track. You can also build your own schedule by creating a free account with Sched.org by selecting "SIGN UP" in the top right corner.
AUTHORS: Kyle Redilla, Michigan State University; Trenton Smith, Missouri Department of Conservation; Barbara Keller, Cervid Program Supervisor, Missouri Department of Conservation; Joshua Millspaugh, University of Montana; Robert Montgomery, Michigan State University
ABSTRACT: Resource selection functions (RSF) have become one of the most popular choices among ecologists for understanding space use from animal telemetry data. Inference is typically desired at the population level, and a common technique is to pool data from all animals in a population and fit a model, where coefficients at the individual level are considered random effects drawn from a population-level distribution. This technique has been shown to be valuable for understanding broad scale selection, but when the focal population may be comprised of various intrinsic categories (e.g. age class) or clustered spatially (e.g. two sub-populations occupying different areas of the landscape), valuable information could be lost by pooling for RSF analyses. We investigated the individual variation in resource selection in a population of elk (Cervus elaphus) introduced into the Missouri Ozarks and monitored between 2011 and 2013. We modeled elk location data collected from Global Positioning System (GPS) collars using a Bayesian discrete choice RSF fit to each individual, and explored results in two ways: in terms of the variability among individual RSF coefficients, and in a model selection approach, comparing the importance of variables in predicting selection. Multivariate analyses revealed patterns among selection coefficients in both magnitude and inclusion in models that varied dramatically across individuals with no obvious clustering by age/sex class. Our work demonstrates that there is likely important ecological variation that can be masked when RSF analyses are aggregated at the population-level. This has implications for identifying the ecological contexts in which routine RSF analyses may flourish or fail. We discuss the implications for our work, highlight the importance of considering resource selection at the level of each individual, and present techniques and considerations for developing individual analyses.
Tuesday February 7, 2017 4:20pm - 4:40pm CST
Arbor I/II
