COMPARISON OF PELLET COUNTS AND AERIAL COUNTS FOR ESTIMATING DENSITY OF MOOSE AT ISLE ROYALE: A PROGRESS REPORT

Abstract

The insular population of moose at Isle Royale offers a unique opportunity to compare the effectiveness of pellet counts vs. aerial counts for estimating moose numbers. In mid-winter, moose were counted every year from the air within small sample plots distributed according to an optimum-allocation stratification. For 8 of the winters during 1980-1993, pellet groups were counted in early spring on clusters of sample plots laid out as a relatively uniform grid. A correlation between estimates from aerial and pellet surveys for all separately analyzed sectors of the island with minimal data showed a significant but very low coefficient of determination, r², of 0.24. This increased to 0.44 when one highly influential, aerial-stratum outlier was removed. Based on the 6 winters for which pellet data were adequate, there was relatively close parallel between the methods of an increasing trend, culminating in the highest density of moose on the island since systematic studies began in 1959. No significant differences were found in regression-line slopes between the two methods when regressed on year, for either the whole island or the East and West sectors analyzed separately. For unexplained reasons, pellet estimates for the East sector were consistently lower than aerial estimates, while for the West they were consistently higher. The defecation rate employed in converting pellet groups to moose density, 20.9 groups/day, had been derived from aerial results in an earlier study; findings in this study do not suggest that a different rate would give a better match. Use of aerial results to calibrate the pellet results renders the two sets of results interdependent. At the same time it is argued that the best means for estimating the defecation rate in a population is from an accurate aerial census. It is concluded that the pellet method provided a reliability similar to that from aerial counts for tracking trends in the Isle Royale moose. Utility of the two methods is compared, and benefits from employing the two in parallel are suggested. Need for improvements in the sampling design and analytical procedures for both methods is indicated.