PRECISION AND BIAS OF AERIAL MOOSE SURVEYS IN NORTHEASTERN MINNESOTA

Abstract

The Minnesota Dept. of Natural Resources (MDNR) has conducted aerial surveys to estimate moose (Alces acles) numbers each year since 1960. A modified version of the survey protocol was developed in Alaska to estimate moose numbers was adopted in 1982. Most numbers and age/sex ratios were estimated from transects flown within a random sample of stratified survey plots. Portions of some survey plots were resurveyed more intensively to determine whether moose were missed in the initial survey. Data from the intensive survey were used to calculate a sightability correction factor (SCF). While more robust than earlier protocols used in Minnesota, this methodology had several shortcomings. Precision of the population estimates (90% C.I.) ranged from ±21 to ±125% and annual changes in moose numbers and bull:cow ratios were not statistically significant. Annual changes in the calf:cow ratios were significant, however. Imprecision of survey estimates was usually the result of incorrect stratification. A single mis-stratified plot dramatically reduced precision on several occasions. Regression analysis indicated that starting date and survey length explained a significant proportion of the variation in population estimates. Moose in the boreal forest shift into conifer cover in mid-winter and become more difficult to observe. The regression results reflected the bias caused by this shift and underscore the importance of starting the survey early and completing the survey in a short time interval.