Rigby, Elizabeth AJohnson, Douglas HAndersen, David E2017-01-192017-01-192014-10-06https://hdl.handle.net/11299/183534Imperfect detectability can complicate analysis of bird survey data. Adjustment methods to account for imperfect detectability exist, but it is not clear how the benefits of these methods compare to their costs. Graduate student Elizabeth Rigby is constructing a computer simulation of bird surveys to evaluate the effects of survey method on survey conclusions. The computer simulation will create simulated birds, then conduct counts of these birds, taking into account realistic parameters of factors known to affect bird counts. This project is currently in the design and coding phase. In addition to the simulation, she conducted a field study of factors affecting detectability of birds in grasslands. The field study assessed the effects of distance to sound source, wind speed and direction, habitat structure and composition, and bird species on the detection of recorded bird songs. Mock surveys with over 9,000 opportunities to detect a recorded bird song were conducted in fall 2011 and 2012 with 4 observers. Detection of recorded songs was treated as a binary variable and analyzed with logistic regression and mixed models. Distance from the observer and an index of wind speed and direction were the strongest covariates to detection. Models used to predict detections of recorded songs performed well, correctly predicting detections 68-90% of the time (depending on species). Observer effects were important; odds of detection for inexperienced observers were only 26% of those of the primary observer. Detection around a sound source was asymmetrical and heavily affected by wind direction. Design and coding for the computer simulation, as well as analysis of field data, will continue in 2014.enReport