RPI Frequently Asked Questions


Q. How does the RPI method account for part-time coverage?


A. It is better to have full coverage, but we can still get good estimates of annual indexes and trends from incomplete data. The statistical procedures we use basically compare the count on any date in a particular year with the available counts on the same dates in all other years. Although we don’t actually estimate the missing or incomplete counts, the effect is like inserting an estimate for the missing data, using all available counts in all years to calculate the most likely value for the missing data.



Q. If my site is missing a year of data, how does this impact the analysis?

A. Although it is better to have data for all years, we can still calculate a trend based on the years that are available. Of course, the trend estimate will be more accurate if all years are represented. One or two missing years in a long set of data will probably have little effect on the trend estimate, but the reliability of the estimate will be strengthened if more years are included and this will be reflected in a stronger statistical significance. Yes, it is better to have data for all years, but missing one or two is not fatal.


Q. How many years of data do I need to enter?


A. RPI has set a lower limit of 10 years to be included in its analyses. This is based on statistical and biological experience. We know that raptor populations vary from year and do not usually follow a steady upward or downward trend. Some go through 3-5 year or longer cycles related to prey populations. Consequently it is difficult or impossible to detect a meaningful trend with short data sets. Therefore we decided not to try to estimate trends with less than 10 years of data.


Q. I have 20 years of data for a single species (e.g. broad-winged hawk) - can these data be analyzed in RPI?


A. Yes, but we tend to give priority to sites with good data for several species. We will likely get around to analyzing single-species sites fairly soon, so please keep entering these data in HawkCount.

Q. How often are the trend maps/graphs updated?


A. Our intent is to do a 2013 update of the trend maps, graphs, and regional assessments, and repeat every other year thereafter.

Q. Entering hourly data is tedious.  Why do you require hourly data for RPI analyses?  Won't daily data do?


A. The problem with daily data is that we don’t know when the hawks were counted. Why is that important? To make consistent comparisons between years, we need to have counts that are standardized as much as possible. We use the hourly data to select counts from specific hours (e.g. 9 a.m. to 4 p.m.) that were covered most consistently and had the highest numbers of the each species. We can’t do this standardization if only daily data are entered. Also, some recent analyses use the actual hourly data to generate indices and trends. Additonally, it is good to have the hourly data in the HawkCount archive for possible future use by other researchers. Sorry about this, but the hourly data really are more useful. If you can’t face entering hourly data now, please continue to enter daily data, and keep your hourly records for possible upgrading in HawkCount later.



Q. My site quits counting on November 15.  Will that give adequate coverage for late season species like Golden Eagles and Rough-legged Hawks?


A. A significant fraction of Golden Eagle migration occurs after November 15, including many of the largest single-day flights in the east. We recommend counting until November 30 at minimum. Several monitoring sites in the Appalachian region continue well into December in large part to document late season eagle movements. Recent telemetry data also shows that some Golden Eagles are still migrating into late December, and there is evidence at some sites of a shift in flight timing towards later dates, perhaps due to warmer conditions at high latitudes. However, keep in mind that access to some sites (esp in the west) may be impossible after early snow/ice storms – personal safety should not be compromised for more complete data!


Rough-legged hawk movements occur primarily after November, and in general this species is very poorly monitored in fall by the existing monitoring network (as evidenced by spring counts at Great Lakes sites being much greater than fall counts). Winter raptor surveys/Christmas Bird Counts are a more effective monitoring tool than fall migration counts for this species.



Q. Can RPI detect whether hawks take alternate routes in different years (e.g. coastal versus mountains)?


A. Raptor migration routes tend to be fairly consistent from year to year over larger spatial scales, i.e. 100s of km; however, routes certainly vary over moderate distances (say 10s of km) as a function of weather conditions. Spatio-temporal analysis of RPI data could possibly be used to detect such shifts in migration routes, but to date this has not been attempted. One could imagine comparing annual indices to 10-yr means across different sites for a particular species and looking for spatial patterns of positive or negative change.


However, such an analysis would be complicated by the inherent variability in annual counts that is due not only to shifts in migration routes between years, but also to (1) actual changes in numbers of raptors, and (2) differences in detectability of migration between years (particularly for species like broad-winged hawks that often use thermal lift to migrate at very high altitude). This inherent variability in count data is why RPI insists on a minimum of 10 years of consistently collected data before determining a trend (see question above).