Sand, H., Zimmerman, B., Wabakken, P., Andrën, H. & Pedersen, H.C. Using GPS technology and GIS cluster analyses to estimate kill rates in wolf ungulate ecosystems. Wildlife Society Bulletin 33, 914-925 (2005).

 The authors used GIS in combination with data from GPS collars to estimate feeding behavior of Grey Wolves in Scandinavia.  This study was used to determine if traditional methods of estimating kill rates from daytime aerial surveys are accurately reflecting prey use by wolves, and if new technology can refine these models to make them more precise.  Traditional models assume 1-2 known locations per day per individual should accurately represent number of large game animals killed by a wolf pack for food.  This study fitted wolves with GPS collars that transmitted a location 1 an hour for the entire study period, day and night.  They buffered these points with 25, 50 and 100 meter buffers, projected them in ArcGIS over a gridded map of the study site, and used overlapping buffers to determine ‘clustering’, each cluster of points marking an individual prey carcass.  These GIS selected sites were then visited by the researchers to examine the effectiveness of the GIS system.  The GPS clustering technique identified about 93% of wolf-killed moose carcasses found in the field if using a 200 m buffer, but only 87% with a 100 m buffer, even with 1 location per hour being recorded by the GPS collar.  The researchers estimate that with the traditional regime of 1 or 2 locations per day, only about 10% of moose kills would be detected.  Also, using clusters recorded during daylight detected about 41% of carcasses, whereas night time clusters detected about 78% of carcasses.  This has serious ramifications for all prey modeling done for wolves using the traditional methods.  This paper showed an excellent way to integrate non-invasive mapping into hypothesis testing and model selection.