How the program works

Formatting the data

Dealing with historic data can be particularly challenging. Protocols often change over time and information recorded under more modern protocols may not have been collected. To insure data are compatible across all years, Data explorer simplifies surveys to 5 min unlimited radius counts and removes all flyovers.

Analyzing the data

The program uses R to analyze overdispersed Poisson linear models which control for time of day, observer, and point-level variation for all species with >10 detections. Models for species with 10 or less detections usually provide biased results; therefore, these species are classified as not having enough data for analysis. Estimates come from a single global linear model. For example, for Morning Doves the R code would be: MODO=glm(count~ year+factor(point)+factor(obs)+hour, family=quasipoisson(),data=MODOData. The complete R code for analysis is available in the file "testBatchPy.R" in the program directory.

Graphing

All graphing is done by sending command line scripts to R.
Comments