Impact of Wolf on livestock in Northern Italy and prediction of predation risk

Mortality due to illegal killing by humans is still today one of the main risk factors for the wolf (Canis lupus) and the effective management of conflict between wolves and livestock breeding is a great concern for species conservation. We assessed the impact of wolf on livestock in an area of Northern Apennines analyzing data provided by the Provincial Administration of Piacenza and additional data directly collected through interviews to farmers during the period 2005-2012. We recorded : i) the number of farms that have suffered predation (classified by : livestock species, productive orientation, rearing method, level of surveillance and preventive methods), ii) the number of predation events and the number of preyed animals. The effects of management factors were evaluated by Likelihood Ratio (exact test with permutation) as regard the frequencies of predation and by Multifactorial Analysis of Variance as regard the average number of heads preyed upon for each event of predation. By Regression and Curve-fit Analyses we evaluated the relationships between the number of predation events and of the number of animals preyed and 21 variables characterizing the grazing areas. Then we formulated a predictive model of the predation probability, through a forward stepwise Binary Logistic Regression Analysis with pasture characteristics as predictors. Livestock species, productive orientation, rearing and preventive methods resulted the most important factors affecting the frequency of predation events and the number of preyed animals. The average number of preyed heads per event increased during the study period and was greater for sheep than other species. The number of predation events showed significant relationship with the pasture slope, the percentage of mixed deciduous forest, the percentage of total deciduous forest and the distance from villages. The number of preyed animals was significantly related with the pasture exposure and with the percentage of mixed deciduous forests. In the model of predation risk entered the rearing methods, the shape index and the percentage of coniferous forest cover, both with a negative effect on the predation probability, whereas exposure had a positive effect. 33.3% of the pastures resulted at low predation risk (Probability of predation = 0.0 - 0.3), 4.2% at medium predation risk (Probability of predation = 0.3 - 0.6), and 62.5% at high predation risk (Probability of predation > 0.6). Overall, 64.6% of the pastures were potentially at risk of predation (Probability of predation > 0.5).