In many biological settings, the construction of a hierarchical model is an effective way to synthesize a complex data set.  With animal population surveys, this model is often constructed to explicitly capture underlying biological characteristics which contribute to the overall variation in the survey data.  There is a tendency, however, for this hierarchical model to become quite complex as the model evolves and additional biological information  is incorporated.  We will discuss the use of Bayesian selection methods to assess the inclusion of additional information and guide us through the model building process.  As an example, we will assess the population trend of Florida manatee, Trichechus manatus latirostris.