Estimating scavenger and search bias for collision fatality surveys of large birds on power lines in the Karoo, South Africa

Abstract

Bird collision mortality associated with power lines is a global conservation challenge, but periodic carcass counts along power lines underestimate actual mortality because of survey biases. We conducted a scavenger trial over 90 days and three observer detection trials after 1, 48 and 90 days using 44 wild goose carcasses to identify explanatory variables and estimate correction factors for these biases. Scavenging activity was high but mostly occurred in situ; only 16% (95% binomial CI 5–27%) of carcasses were removed by the end of the 90-day trial period (all within the first three weeks). Observer experience did not significantly affect detection rate, with experienced observers locating 75% (95% CI 68–83%) of carcasses compared with 67% (95% CI 58–75%) by naïve observers. Vegetation height, distance from the line and carcass spread were significant explanatory variables. The pattern in detection rate linked to carcass age was variable; there were fewer detections after 48 days, but rates at day 1 and day 90 were similar for experienced observers because bleached bones were highly visible. This suggests that long survey intervals are acceptable for the detection of large bird carcasses in arid environments such as the Karoo where collisions are infrequent but widespread events.