Effects of GPS Sampling Intensity on Home Range Analyses

The two most common methods for determining home ranges, minimum convex polygon (MCP) and kernel analyses, can be affected by sampling intensity. Despite prior research, it remains unclear how high-intensity sampling regimes affect home range estimations. We used datasets from 14 GPS-collared, white-tailed deer (Odocoileus virginianus) to describe the size and location accuracy of home range estimates calculated from different sampling regimes. We compared monthly home range estimates from seven sub-samples (480, 360, 180, 90, 60, 30, and 15 locations) to the home range estimates of the complete datasets (720 locations). Minimum convex polygon (MCP) home range size estimates calculated from datasets with > 180 locations were not statistically different. Areas calculated with 60-90 locations may underestimate MCP size by 50% or more. As demonstrated in past studies, we found that kernel home range analyses accurately estimated home range size for all sampling regimes. However, considerable locational errors were associated with lower sampling regimes, resulting in misclassifications of areas of use and non-use. An average locational error > 40% was observed for our least intensive sampling regime, while sampling regimes collecting 480 and 360 locations had less than 10% relative error. Since GPS technology can generate large sample sizes, researchers should use kernel analyses because MCP ignores much of the data generated. Also, because significant location error may be associated with MCP home ranges calculated from small sample sizes, the results of many previously published studies should be interpreted with care.

Starting page
13
Ending page
17
ID
77112