The ability to detect trends in population abundance is of obvious interest to wildlife managers. In recent years, the probability of detecting defined population trends has been the most common method of assessing monitoring programs. Such analyses require many assumptions, including a model for population change and a model for variance. To demonstrate potential effects of these assumptions on power analysis results, we present data for Red-tailed Tropicbirds (Phaethon rubricauda) from Tern Island, Hawaii. Depending on our assumptions, the power to detect a 50% decline over 10 years varied from 80% to 100%. We argue that monitoring standards based upon the ability to detect population trends should be applied cautiously. As a complementary approach, we propose that monitoring standards should emphasize attributes of sampling design that increase precision (e.g., randomization, bias, and detection probability). By using standards of precision, managers can focus on the sources of variation that can be minimized. A sampling design approach to monitoring standards provides a useful complement to standards of statistical power to detect annual trends.
Additional publication details
Is statistical power to detect trends a good assessment of population monitoring?