An article published in November of this year by Wilson et al.  of The Biodiversity Consultancy presents a review of raptor carcass persistence trials and the practical implications for fatality estimation at wind farms.

The site dependence of hazards presented to birds by turbines needs to be captured in data to inform the planning and siting process for future wind energy projects towards the minimisation of bird casualties.

Each candidate wind turbine site will vary in terms of the number of birds and flights of birds through the area. It will also vary in the duration of time for which carcasses of fatally injured birds persist on site, which is influenced by factors such as removal by scavenger species or decomposition between searches.

To improve the estimation of the number of fatalities caused by an operational wind energy project, two additional studies should be conducted post-construction: a fatality search around randomised samples of turbines; and the carcass persistence experiment, whereby the set of habitats in the zone of any fallen birds around the turbines should be represented by areas chosen elsewhere. At these sample points, carcasses should be placed and monitored to determine their persistence.

After the windfarm construction, a dataset was collated as records of single raptor carcasses. For each carcass record, the variables used in the persistence analysis were: the last time the carcass was known to be present and the first time it was recorded as absent. The GenEst package within R was used to analyse the data. The contributions to variation in persistence arising from differences in the effectiveness of carcass searches and the region were assessed respectively by ‘trial’ and ‘biome’, two explanatory variables incorporated in GenEst.

There were 17 study sites incorporated in the analysis split across four biomes: Temperate Grasslands, Savannah/Shrubland, Temperate Broadleaf/Mixed Forests, Temperate Conifer Forests, and Deserts and Xeric Shrublands. Two of the 17 study sites had multiple trials across the dataset, with a total of 22 trials. Most carcass persistence rates pertained to landfowl, with raptor records derived from only two trials.

The model was initially used to estimate carcass persistence across all sites, which quantifies the rate of degradation of the carcass in the environment, including the utilisation by other species. The model was then run with the trial and the biome as covariates.

Carcass persistence estimates were obtained as medians for specific trial sites. Across the 17 trials, there was no significant correlation between the number of carcases in the trial and the median carcass persistence.

In addition, the GenEst package was used to produce fatality estimates by generating a mock dataset having the calculated median persistence value for all sites and median persistence values for trial datasets.

A key result is that carcass persistence time showed significant variation between different trials, but not between biomes. However, it is emphasised that four of the world’s 14 land biomes is probably an insufficient sample and that in biomes such as the African savannah and tropical areas with high humidity or rainfall, carcass persistence may vary significantly within the biome.

Overall, carcass persistence rates between trials varied substantially but had only a minor effect on the fatalities estimated from the model when both 14-day and 28-day search frequencies were simulated since carcass persistence time was found or estimated to be hundreds of days in most trials.


Wilson D, Hulka S, Bennun L. 2022. A review of raptor carcass persistence trials and the practical implications for fatality estimation at wind farms. PeerJ 10:e14163

Link to full article:

A review of raptor carcass persistence trials and the practical implications for fatality estimation at wind farms [PeerJ]