Are kernels the mustard? Data from global positioning system (GPS) collars suggests problems for kernel home- range analyses with least-squares cross-validation

1. Kernel-density estimation (KDE) is one of the most widely used home-range estimators in ecology. The recommended implementation uses least squares cross-validation (LSCV) to calculate the smoothing factor (h) which has a considerable influence on the home-range estimate. 2. We tested the performance of least squares cross-validated kernel-density estimation (LSCV KDE) using data from global positioning system (GPS)-collared lions subsampled to simulate the effects of hypothetical radio-tracking strategies. 3. LSCV produced variable results and a 7% failure rate for fewer than 100 locations (H = 2069) and a 61% failure rate above 100 points (n = 1220). Patterns of failure and variation were not consistent among lions, reflecting different individual space use patterns. 4. Intensive use of core areas and site fidelity by animals caused LSCV to fail more often than anticipated from studies that used computer-simulated data. 5. LSCV failures at large sample sizes and variation at small sample sizes, limits the applicability of LSCV KDE to fewer situations than the literature suggests, and casts doubts over the method's reliability and comparability as a home-range estimator. © 2005 British Ecological Society.

[1]  C. Loader.  Local Likelihood Density Estimation , 1996 .

[2]  Nicholas J. Aebischer,et al.  Compositional Analysis of Habitat Use From Animal Radio-Tracking Data , 1993 .

[3]  Robert A. Gitzen,et al.  Analysis of Animal Space Use and Movements , 2001 .

[4]  W. D. Spencer,et al.  An evaluation of the harmonic mean measure for defining carnivore activity areas , 1984 .

[5]  Robert E. Kenward,et al.  RANGES V: an analysis system for biological location data , 1996 .

[6]  J. L. Gittleman,et al.  A Common Rule for the Scaling of Carnivore Density , 2002, Science.

[7]  D. W. Scott,et al.  Cross-Validation of Multivariate Densities , 1994 .

[8]  K. Dixon,et al.  Harmonic mean measure of animal activity areas , 1980 .

[9]  J. Brashares.  Ecological, Behavioral, and Life‐History Correlates of Mammal Extinctions in West Africa , 2003 .

[10]  R. Bonduriansky,et al.  Eliminating autocorrelation reduces biological relevance of home range estimates , 1999 .

[11]  M. Wand.  Local Regression and Likelihood , 2001 .

[12]  B. Worton.  Using Monte Carlo simulation to evaluate kernel-based home range estimators , 1995 .

[13]  B. Worton.  Kernel methods for estimating the utilization distribution in home-range studies , 1989 .

[14]  M. C. Jones,et al.  A Brief Survey of Bandwidth Selection for Density Estimation , 1996 .

[15]  Kathy H. Hodder,et al.  DENSITY AND LINKAGE ESTIMATORS OF HOME RANGE: NEAREST‐NEIGHBOR CLUSTERING DEFINES MULTINUCLEAR CORES , 2001 .

[16]  R. Powell,et al.  An Evaluation of the Accuracy of Kernel Density Estimators for Home Range Analysis , 1996 .

[17]  Gary C. White,et al.  A Comparison of Home-Range Estimators Using Monte Carlo Simulation , 1990 .

[18]  Joshua J. Millspaugh,et al.  Comparison of least-squares cross-validation bandwidth options for kernel home-range estimation , 2003 .

[19]  E. Debevec,et al.  LINEAR HOME RANGES: EFFECTS OF SMOOTHING, SAMPLE SIZE, AND AUTOCORRELATION ON KERNEL ESTIMATES , 2001 .

[20]  R. Courtois,et al.  Effects of sampling effort based on GPS telemetry on home-range size estimations , 2002 .

[21]  B. Silverman.  Density estimation for statistics and data analysis , 1986 .

[22]  Joshua J. Millspaugh,et al.  Effects of sample size on kernel home range estimates , 1999 .

[23]  J. Ginsberg,et al.  Ranging behaviour and vulnerability to extinction in carnivores , 2000 .