Improvement of errors in radiotelemetry locations of brown bears, Ursus arctos, in Hokkaido, Japan

Abstract

The errors and the sources of errors made while fixing radiotelemetry locations were estimated in two brown bear, Ursus arctos yesoensis, study areas, in the Shiretoko and Oshima regions of Hokkaido. We measured the sampling error and the bias of test transmitters placed at various points in our study areas. The means of our sampling errors were approximately equal to those previously reported by Springer (1979), while the maxima of our sampling errors and the means of our biases were larger than those of Springer's (1979). We also assessed the amount of error in estimating locations based on measuring three bearings. As the sizes of triangles were positively related to the degree of error in estimating their points, we excluded large (>6.25 ha) triangles from the analysis. The maximum values of the 99% confidence intervals for normalized error distances were 321.4 m in the Shiretoko area and 302.3 m in the Oshima area. These values were compared with Lenth's (1981) Andrews estimators calculated from the original data sets of both study areas. The direct method for estimating radiotelemetry error which we used in this study is easy to calculate and proved not to be inferior to Lenth's (1981) method.