We describe a method for surveying birds in river habitats that allows for rapid assessment across broad spatial scales and estimation of detection probabilities. Our river survey approach incorporates a double-survey technique, whereby observers in two canoes simultaneously survey birds along a river reach. Data are in the form of a two-sample mark-recapture history, and covariates suspected of influencing detection probabilities can be included in the modeling process and evaluated using information-theoretic approaches. We provide an example using the method along the Madison and upper Missouri rivers in Montana. Overall, detection probabilities for each observer ranged from 57% to 89%, and combined detection probabilities (the likelihood of at least one observer detecting an individual) were consistently high (88% across all species). Detection probabilities across species were positively correlated with body mass. Detection probabilities for some species were influenced by observer, river conditions, and whether species were in groups or alone; groups were more detectable, and individuals in slow-flowing and wide sections of river were more detectable. Boat-based double surveys are a viable method for estimating detection probabilities of birds in river habitats, and double surveys should be considered in other aquatic systems.
© 2006, University Of California Press. View original published article in JSTOR.