Occupancy surveys for inconspicuous species like the American pika (Ochotona princeps) often rely on indirect sign — scat, haypiles, calls — that varies widely in how reliably it indicates true presence. Standard occupancy models typically pool all of these sign types into a single detection process, ignoring the fact that fresh scat, old haypiles, and calls carry very different risks of being missed or misidentified. This study develops a “multi-sign” occupancy model that instead estimates a separate detection process for each sign type, including its own false-positive rate, and applies it to over a decade of pika surveys across five Pacific Northwest parks.
Modeling detection this way mattered: simplified versions of the observation process (assuming perfect detection, or ignoring false positives) tended to inflate occupancy estimates and change which environmental variables appeared to drive occupancy trends. Notably, forb cover had a much stronger estimated effect on occupancy dynamics in the full multi-sign model than in the simpler models — a difference that could mislead management decisions if the wrong model were trusted.
Citation: Goldman, M.R., Shinderman, M., Jeffress, M.R., Rodhouse, T.J., Shoemaker, K.T. 2023. Integrating multiple sign types to improve occupancy estimation for inconspicuous species. Ecology and Evolution 13(5):e10019. https://doi.org/10.1002/ece3.10019