Human related mortality is a major threat for large carnivores all over the world and there is increasing evidence that large predators respond to human related risks in a similar way as prey respond to predation risk. This insight recently led to the conceptual development of a landscape of coexistence that can be used to identify areas which can sustain large predator populations in human dominated landscapes. In this study we applied the landscape of coexistence concept to a large predator in Europe. We investigated to what extent Eurasian lynx (Lynx lynx) habitat selection is affected by human disturbance in a human dominated landscape. More specifically, we were interested in the existence of a tradeoff between the availability of roe deer, one of their main prey and avoidance of human disturbance and how this affects the spatio-temporal space use patterns of lynx. We found that lynx face a tradeoff between high prey availability and avoidance of human disturbance and that they respond to this by using areas of high prey availability (but also high human disturbance) during the night when human activity is low. Furthermore our analysis showed that lynx increase their travelling speed and remain more in cover when they are close to areas of high human disturbance. Despite clear behavioral adjustments in response to human presence, prey availability still proved to be the most important predictor of lynx occurrence at small spatial scale, whereas human disturbance was considerably less important. The results of our study demonstrate how spatio-temporal adaptations in habitat selection enable large carnivores to persist in human dominated landscapes and demonstrate the usefulness of the concept of a landscape of coexistence to develop adaptive management plans for endangered populations of large carnivores.
deer_data
This table contains the GPS locations of roe deer associated with habitat variables and temporal variables that were used to build a habitat model (RSF) for roe deer. The table is divided into used (actual) deer locations and random locations (loc_id=NA) in a ratio of 1:10 (column "use"). The table also includes an animal id. Swisstopo in the column headers refers to the source of the environmental variables. Cover swisstopo is a dummy variable for open/cover. Slope_sq and altitude_sq are the squared slope and altitude variables. Aspect swisstopoS is the southern exposition. Hum_indx is a composite of road_dist and house density. The 8 temporal variables are time harmonics of a Fourier transform for time of day (tsin, tcos, tsin2, tcos2; period of 24) and day of year(ytsin, ytcos, ytsin2, ytcos2; period of 365) All continuous variables in this table are mean centered and standardized to a SD=1. For information on the raw data or for R-code used to run the models please contact the data owner.
lynx_data
This table contains lynx steps associated with habitat variables and temporal variables that were used to build a step selection function model for lynx. The table is divided into used (actual) lynx steps and random steps in a ratio of 1:10 (column "use"). The random steps match to their corresponding location id in the used steps. Swisstopo in the column headers refers to the source of the environmental variables. Hum_indx is a composite of road_dist and house density. Prey_avail are the values of a deer RSF, which was used as a proxy for prey availability. Dist2 refers to the step length (distance between the previous and current location in meters). The 8 temporal variables are time harmonics of a Fourier transform for time of day (tsin, tcos, tsin2, tcos2; period of 24) and day of year(ytsin, ytcos, ytsin2, ytcos2; period of 365). All habitat variables were measured at the beginning of a step except for “hum_indx_loc2”, which was measured at the beginning of a step. All continuous variables in this table are mean centered and standardized to a SD=1. For information on the raw data or for R-code used to run the models please contact the data owner.
lynx_data_strata_cluster_11112020
This updated table contains two additional variables that were missing in the first lynx_data table. Namely, this is 'loc_id' which isused for the strata argument in the model specification and 'id_anim' which is the individual lynx id for implementing the two-step approach. The other data is identical to the 'lynx_data' table.