The goals of this study were to investigate the possibility of using kernel techniques to estimate male breeding territory size and delineate core areas, focusing on a small nontransmitter bearing bird, the cerulean warbler. We then compared the performance of kernel estimators with traditionally used minimum convex polygons (MCP).
Given the lack of a consistent across‐male sample size–area relationship, we opted to use each male's full set of locations in the kernel calculation rather than standardizing sample size across males.
All locations collected for each male were biologically independent though statistically autocorrelated. Subsampling locations did not achieve independence even at time intervals far exceeding biological independence.
The physical space bounded by kernel and MCP methods differed drastically in certain cases, especially in situations where there were large areas within a territory that were never visited during our data collection sessions.
Kernel methods of territory estimation were far more accurate and informative than MCP for cerulean warblers. We suggest that evenly sampling individuals in a biologically relevant manner during a strictly defined study period is more important than standardizing sample size across individuals. Furthermore, sampling regimes can safely be guided by biological vs. statistical independence timelines.
Avian biologists should consider kernel estimators as an option especially for habitat selection studies where accurate territory boundary and size estimation is crucial.