Field trials of line-transect surveys of bird carcasses in agro-ecosystems of Argentina’s Pampas region

Abstract

Pesticides caused high mortality of birds in agro‐ecosystems of Argentina's Pampas region during 1995–2003. As part of a program to monitor mortality events, we trained observers in distance sampling methods, conducted field trials under simulated conditions to assess the degree to which the critical assumptions of line‐transect surveys were met, and estimated the density and number of chicken carcasses randomly placed along fixed transects in corn stubble, corn, alfalfa, wheat, pasture, and forest. The assumption of 100% detection probability of carcasses at 0 distance was met. We measured cluster size (1–5 carcasses) exactly and measured perpendicular distance from a cluster to transect centerline (0–40 m) with little error (absolute difference between actual and measured distances: x̄=0.14 m, SE=0.01, n= 121). Cluster detection was not size‐biased in corn stubble but was size‐biased in corn, alfalfa, wheat, pasture, forest, and all strata combined. Thirteen 3‐observer teams performed equally well in field trials regardless of having or not having observers with previous experience conducting transect surveys of bird carcasses in agro‐ecosystems. Teams with experienced and inexperienced observers missed clusters near the centerline (defined here as 1.5–5 m for wheat and 5–15 m for other strata) but detected all clusters on the centerline (defined here as 0–1.5 m for wheat and 0–5 m for other strata). Percent relative bias was 0.36% for corn stubble (n=39); 5.63% for the combined data of corn, alfalfa, wheat, pasture, and forest (n=73); and 0.51% for all strata (n=121). Bias estimates were small in relation to standard errors of density estimates, and the density estimator produced 95% confidence intervals that always included true densities. Predators and scavengers started removing carcasses during the first day of exposure (cumulative survival: x̄=0.69, SE=0.04, n=36) and stopped removing them after day 5 (cumulative survival: x̄=0.04, SE=0.02, n=5). Carcass removal rate averaged −0.38/day (SE=0.03). Although line‐transect sampling represents a viable method to estimate the density and number of dead birds in agro‐ecosystems, scavenging data should be collected to adjust these estimates, even when surveys are conducted the day following pesticide application. An effective communication network between collaborators and the general public and observer readiness are needed as part of research and monitoring efforts to guide decision‐making on agricultural practices, promote law enforcement, and manage wildlife in agro‐ecosystems.

Estimating carcass persistence and scavenging bias in a human-influenced landscape in western Alaska

“…Overall, daily carcass persistence in the tundra habitats of our study area was low. In most cases, fewer than 50% of the carcasses were likely to persist for 24 h. Our estimates are considerably lower than published estimates of bird carcass persistence in terrestrial habitats (Tobin and Dolbeer 1990, Linz et al 1991, Kostecke et al 2001, Rivera‐Milan et al 2004). However, our persistence rates were higher than observed for small carcasses placed in fields in Saskatchewan, Canada (Wobeser and Wobeser 1992).…”

Factors influencing experimental estimation of scavenger removal and observer detection in bird–window collision surveys

“…This bias has been assessed for many mortality sources, and studies show that carcass persistence varies with season, carcass size, carcass age (time since carcass deposition), local vegetation, and the surrounding landscape (Rivera‐Milán et al. , Prosser et al. , Ponce et al.…”

“…), but experience had no effect in many field tests (Rivera‐Milán et al. , Stevens et al. , Schutgens et al.…”

Modeling carcass removal time for avian mortality assessment in wind farms using survival analysis

“…These authors also estimate the percentage of unscavenged carcasses by dividing the number of carcasses remaining at the end of a time period by the number of carcasses at the beginning of the period. Rivera-Milan et al (2004) and Kostecke et al (2001) used survival analysis to estimate removal rates of bird carcasses in non wind farm related contexts. These authors used parametric survival analysis methods, modeling data through an exponential regression.…”