We received more than 15,000 reports of surveys conducted from 1986 to 2005. We reviewed all survey forms to ensure that the number of eagles reported in each age class tracked with the total number reported. We excluded surveys from the analysis if: (1) the count was conducted before 1 Jan or after 25 Jan; (2) the route covered was inconsistent from year to year; (3) the route was not described clearly enough to determine if it was similar to other years; (4) the survey method was not listed; or (5) we were unable to obtain information on size of the survey route. We tried to verify that route coverage was consistent from year to year by comparing reported mileages among years. In many cases, we contacted observers and state coordinators to verify starting and ending points as well as total area surveyed. We corrected or eliminated records when we became aware of cases where observers covered areas less than or greater than the defined standard route. We only used data from routes that were surveyed consistently in at least 4 years where at least 4 eagles were counted in a single year. Survey routes with <4 eagle observations/year are probably in marginal or unsuitable wintering habitat; including data from routes where few or no eagles were ever seen would have biased our trend estimates towards zero. We evaluated count data to determine if fog, precipitation, or changing transportation methods might have affected the ability of observers to detect eagles and thereby influence trend estimates. Based on our evaluation, we excluded surveys that used a different transportation method on a particular route in a minority of years, but we included surveys conducted during fog or precipitation. After all exclusions, our final data set includes 178,896 observations of eagles during 8,674 surveys of 746 routes in 43 states.
Our analyses indicated that we could not use a survey route in the trend analysis if the method of survey (air, ground, water) changed. So if we have 10 years of aerial survey on a given route and 1 year of ground survey, we have to throw out the ground survey. However, if we have at least 4 years of aerial surveys and at least 4 years of ground surveys (different years than the aerial surveys), we have considered them as separate routes and use both of them in the analysis.