Efficacy of Northern Goshawk Broadcast Surveys in Washington State

Statewide surveys conducted for northern goshawks (Accipiter gentilis) by the Washington Department of Fish and Wildlife (WDFW) in the early 1990s had relatively low detection rates throughout the state and different rates of detection between eastern (1bird/55 km) and western Washington (1 bird/174 km). To investigate the possibility that survey methods or regional habitat characteristics affected detection rates, we tested broadcast calls at 40 nest stands known to be occupied by nesting northern goshawks. During 439 station visits and 210 trials, we recorded 109 detections, 68% of which were vocalizations. Northern goshawks at 37 of the 40 (93%) occupied territories were detected at least once, including 4 occupied sites that failed. Detectability was greater at successful nests by 37% during the postfledging period and accounted for 87% of the responses. Logistic regression modeling identified distance of the surveyor to the nest as the only factor correlated with detection rates (P < 0.001). Probability of detecting northern goshawks at occupied stands increased as nest were approached from 400 m (0.20), 250 m (0.25), and 100 m (0.42). Binomial expansion of detection probabilities at a single nest visit found broadcasting to attain >=90% detections required 5 visits at 100 m, 5 at 250 m, and 10 at 400 m. Because analysis of detection rates by area did not show an effect of vegetation screening, slope, or topography, differences in the relative abundance of northern goshawks from earlier surveys reflect true variation in abundance, differences in nest success, or some combination of both. Results lend more justification for use of broadcast calling for northern goshawks in Pacific Northwest forests and provide a hypothetical model for estimating survey costs as a function of detection probability, trial frequency and station spacing.