Diurnal Human Activity and Introduced Species Affect Occurrence of Carnivores in a Human-Dominated Landscape

Abstract

Diurnal human activity and domestic dogs in agro-forestry mosaics should theoretically modify the diurnal habitat use patterns of native carnivores, with these effects being scale-dependent. We combined intensive camera trapping data with Bayesian occurrence probability models to evaluate both diurnal and nocturnal patterns of space use by carnivores in a mosaic of land-use types in southern Chile. A total of eight carnivores species were recorded, including human-introduced dogs. During the day the most frequently detected species were the culpeo fox and the cougar. Conversely, during the night, the kodkod and chilla fox were the most detected species. The best supported models showed that native carnivores responded differently to landscape attributes and dogs depending on both the time of day as well as the spatial scale of landscape attributes. The positive effect of native forest cover at 250 m and 500 m radius buffers was stronger during the night for the Darwin's fox and cougar. Road density at 250 m scale negatively affected the diurnal occurrence of Darwin´s fox, whereas at 500 m scale roads had a stronger negative effect on the diurnal occurrence of Darwin´s foxes and cougars. A positive effect of road density on dog occurrence was evidenced during both night and day. Patch size had a positive effect on cougar occurrence during night whereas it affected negatively the occurrence of culpeo foxes and skunks during day. Dog occurrence had a negative effect on Darwin's fox occurrence during day-time and night-time, whereas its negative effect on the occurrence of cougar was evidenced only during day-time. Carnivore occurrences were not influenced by the proximity to a conservation area. Our results provided support for the hypothesis that diurnal changes to carnivore occurrence were associated with human and dog activity. Landscape planning in our study area should be focused in reducing both the levels of diurnal human activity in native forest remnants and the dispersion rates of dogs into these habitats.

Fig 1. Map of the study area characterized by a human-dominated land-use mosaic surrounding the Nabuelbuta National Park.

Fig 2. Contour plots showing model-predicted occurrence probabilities (ψ) of Darwin’s fox (A and B), culpeo fox (C and D) and cougar (E and F) as function of two landscape—scale covariates: road density at two different scales (250m radius buffer and 500m radius buffer) and the amount (%) of native forest at 500m.

Red isolines indicate combinations of the two covariates predicting a particular (ψ) level (with the blue isocline showing ψ = 0.5). Vertical and horizontal dashed lines indicate the mean value of the covariate, as measured in the study landscape.