Accounting for incomplete detection: What are we estimating and how might it affect long-term passerine monitoring programs?

A primary objective of ecological monitoring programs typically includes the efficient detection of population trends. Passerines as a group are important ecological indicators and are often included in such programs to provide information on multiple species with a single survey technique. However, commonly used field and analytical approaches may not provide appropriate inference or sensitivity due to assumption violations and differences in the proportion of the population exposed to sampling. Recent methodological developments utilizing repeated point counts and an N-mixture modeling approach for analysis may produce more consistent and interpretable estimates applicable to the superpopulation of individuals using a site during the breeding season. These estimates should be more useful for monitoring because they are not conditioned on presence or availability as are most single-visit approaches. We used repeated count data collected in Denali National Park and Preserve, Alaska (Denali) between 1995 and 2009 from 12 common passerine species to assess variation in presence and availability throughout the season, estimate trends in superpopulation abundance, and provide recommendations for the design of future monitoring programs. We found that variation in detection due to presence and availability was large and differed among species. After accounting for these sources of variation, we estimated abundance of Wilson's warblers (Wilsonia pusilla) had declined by approximately 48% and fox sparrow (Passerella iliaca) abundance had increased by approximately 250% over 15 years. Combined, our results suggest that if trend estimation is a priority, passerine monitoring programs should formally address all components of the detection process, including the probabilities of presence and availability.