Predation and habitat deterioration are the main reasons for the strong decline of ground-nesting farmland birds such as the grey partridge Perdix perdix in Europe. Grey partridge nests and incubating females are especially vulnerable to predation. We have previously demonstrated that predator activity is much lower inside flower blocks (agri-environment schemes sown with a flower seed mix) than at their edges and that predator activity in flower blocks depends on the surrounding landscape. Here, we investigate whether these differences in predator activity translate into differences in grey partridge nest predation and assess predation patterns relative to landscape and nest site characteristics. We recorded the success of 56 nests of radio-tagged grey partridges between 2009–2017 in an agricultural landscape in Central Germany. We used Bayesian logistic regression to analyse the effects of nest site and landscape characteristics on nest predation on a subset of 46 nests (21 nests successful, 25 predated). Distance to the edge of the nesting habitat was the most important predictor, reducing predation probability from 66.8% at the edge to 18.5% at 85.5 m. Predation probability decreased with increasing length of habitat borders, habitat diversity and the area of permanent grasslands and fallows. Predation probability was higher further from settlements and increased with increasing woodland area in the agricultural matrix. When considering linear landscape structures, nest predation patterns matched the patterns of predator activity from our previous studies. Results suggest that the distance to the edge of the nesting habitat is most important, and that nest predation may be reduced by providing sufficiently broad nesting habitats. Nest predation may further be minimized by increasing habitat diversity and coverage of extensive vegetation types and by establishing conservation measures for grey partridges further away from woodlands. These measures may also benefit other ground-nesting farmland birds. 

Please see the corresponding article for information on data collection and analysis.

R and JAGS are required to run the code. Additionally the packages jagsUI, dplyr, lubridate and loo are necessary to run the code.