Spatial bias in dietary studies can limit our understanding of the feeding ecology of large carnivores

Abstract

Many large carnivores have broad geographical ranges, encompassing ecosystems with a different prey base. Our understanding of their diet could therefore be biased by the spatial concentration of dietary studies into few areas. We propose a protocol to divide the geographical range of large carnivores, into areas that are homogeneous with respect to available food sources, by using the grey wolf (Canis lupus) in Italy, as a case study. We mapped the potential maximum distribution of wolves, on a 10 km grid (n = 2,497), and then performed cluster analysis to classify cells according to their: i) abundance of domestic and wild ungulates, ii) suitability for the coypu (Myocastor coypus) and iii) landscape anthropization. Finally, we checked the percentage of cells in each cluster that were covered by dietary studies in 2007-2013, 2014-2018 and 2019-2023. The distribution range of wolves in Italy can be divided into 5 areas, characterized by different food sources but also by a different spatial coverage from dietary studies. The Alps and some sectors of the Apennines, with low anthropization and abundant wild ungulates, were oversampled. More anthropized areas in Central and Southern Italy, rich in sheep and wild ungulates, as well as anthropized lowlands, with abundant food waste and coypu, were undersampled. Finally, no study was carried out in intensive farming districts of Northern Italy. Our protocol indicates that future studies about the diet of wolves in Italy should focus on anthropized landscapes. There, the consumption of pets could trigger wolf persecution and pathogen transmission, and predation on coypu and the consumption of food waste could increase the exposure to toxic compounds. More broadly, our protocol can improve our understanding about the feeding ecology of large carnivores, as it can be used to: i) assess and put into perspective meta-analytic findings, ii) identify knowledge gaps arising from spatial bias and prioritize new studies in undersampled areas and iii) design sampling schemes for large-scale research.