Research overview: bandwidth selection for kernel home-range estimation

The greatest source of variability in kernel home-range estimates is the smoothing bandwidth, and this overview surveys the competing approaches and their pitfalls. The reference bandwidth tends to oversmooth and inflate range area, least-squares cross-validation often fails or undersmooths with clustered fixes, and plug-in and likelihood-based selectors offer alternatives with their own assumptions. Because bandwidth choice can change estimated area several-fold, comparisons across studies that used different selectors are fraught. The overview explains why this seemingly technical decision has substantive ecological consequences for inferences about cat space use and why the autocorrelation-aware methods that select smoothing from the movement process emerged to resolve the impasse. Understanding bandwidth is essential to reading the older felid home-range literature critically.