Effects of population variability on the accuracy of detection probability estimates

Observing a constant fraction of the population over time, locations, or
species is virtually impossible. Hence, quantifying this proportion (i.e.
detection probability) is an important task in quantitative population
ecology. In this study we determined, via computer simulations, the effect
of population characteristics on estimates of detection probability.
Simulation results showed a consistent and significant over-prediction
of detection probability across sampling methodologies and estimators.
Comparisons between estimators and sampling methods showed significant
differences amongst them (estimators accounting for heterogeneity
are the most accurate). Population size was the most important
factor influencing the uncertainty and accuracy of the estimates. Our
results show the need to include population characteristics (either as a
correction function or as factors in the observation model) when estimating
detection probabilities. The study concludes by suggesting two
methods for using population characteristics to estimate detection
probabilities.