Fitting Population Models to Multiple Sources of Observed Data

The use of population models based on several sources of data to set harvest levels is a standard procedure most western states use for management of mule deer (Odocoileus hemionus), elk (Cervus elaphus), and other game populations. We present a model-fitting procedure to estimate model parameters from multiple sources of observed data using weighted least squares and model selection based on Akaike's Information Criterion. The procedure is relatively simple to implement with modern spreadsheet software. We illustrate such an implementation using an example mule deer population. Typical data required include age and sex ratios, antlered and antlerless harvest, and population size. Estimates of young and adult survival are highly desirable. Although annual estimates are desirable, the procedure also can be applied-with less precision-to data sets with missing values in any of the data series. The model-fitting procedure adjusts input estimates and provides estimates of unobserved parameters to achieve the best overall fit of the model to observed data. Rigorous, objective procedures such as those described here are required as a basis for wildlife management decisions because diverse stakeholder groups are increasing the intensity with which they scrutinize such management decisions.