The wolf is back! Non-consumptive effects of the return of a large carnivore on the use of supplementary feeding sites by roe deer
Data files
Aug 22, 2025 version files 441.89 KB
-
Capreolus_last.csv
356.56 KB
-
Code_All_Analyses.R
38.30 KB
-
Occurrence3.csv
45.63 KB
-
README.md
1.40 KB
Abstract
Understanding how prey species trade-off predation risk and resource acquisition is particularly important for advancing our knowledge of predator-prey relationships. We investigated this by studying the use of concentrated anthropogenic resources, namely supplementary feeding sites, by roe deer Capreolus capreolus before and after grey wolf Canis lupus recolonisation in an area of the eastern Italian Alps. We used camera traps to monitor roe deer visits to feeding sites, where ad libitum food was provided, before and after wolf recolonisation, in winter and spring, to control for seasonal effects. First, we compared the daily cycle of visits using circular statistics. We then used generalised linear mixed models to model the occurrence, duration of visits, and tendency to congregate at feeding sites as a function of wolf presence and season. Roe deer became more diurnal after wolf recolonisation, particularly in winter, while in spring they tended to concentrate their visits around dusk and dawn. Roe deer visits to feeding sites decreased from winter to spring, but only after wolf recolonisation, while their duration was shorter in spring when wolves were absent than in any other period. Roe deer grouping at feeding sites decreased from winter to spring, especially after wolf recolonisation. These results show that roe deer have changed their resource use behaviour since the return of the wolf, adopting a range of behavioural tactics that could mitigate predation risk, while maintaining resource acquisition when more profitable. The increase in diurnality may reduce the temporal overlap with wolves’ predominantly nocturnal activity; access to the resource-rich, but fairly exposed sites mainly occurred during the most limiting season, or with solitary visits. We call for further research to understand whether other unmeasured processes contribute to shaping the observed patterns, such as demographic decline and fine-scale behavioural adjustments (e.g. increased vigilance).
The data consist of 2 tables with data and 1 R code that reproduces all analyses (Code_All_Analyses.R).
Dataset 1 (Capreolus_last.csv)
- Id Event: Id of an event of camera trapping (integer)
- FS: feeding site ID (integer)
- Period: Boolean (spring/winter)
- Year: year of the event, can be either yyyy for spring periods or yyyy/yyyy for winter periods across two subsequent years
- Presence_wolf: presence of wolf in a given period
- maxInd: maximum number of individual detected in an event
- eventStart: timestamp of the first picture of a given event
- eventEnd: timestamp of the last picture of a given event
- eventDuration: duration I seconds of the event (difference btw event end and event start)
- Male: number of males in the event
- Female: number of females in the event
- Unknown: number of animals in the event with sex not identified
- Per_Wolf: combination of season and presence of wolf (used in the models)
Dataset 2 (Occurrence3.csv)
- FS: feeding site ID (integer)
- Day_Fact: day of the year
- Visits: number of daily visits at any FS (events)
- Year: year of the event
- Season: 1 (spring) or 0 (winter)
- Presence_wolf: presence of wolf in a given period
- Per_Wolf: combination of season and presence of wolf (used in the models)
