Air temperature (°C), solar radiation (W/m²), wind speed (m/s) and wind direction (cosine-transformed to range between -1 with wind blowing from the South and +1 with wind blowing from the North) were recorded every hour by the closest weather station (Pont Station, 45°31′36.62N, 7°12′03.36 E; 1951 m a.s.l.; Regione Valle d’Aosta, official data).
Fine-scale temperatures were recorded hourly in the Levionaz Valley using temperature loggers (iButton DS1922L, Maxim Integrated, n = 15 in 2010, n = 17 in 2011) stratified by elevation and hydro-geographic sectors corresponding to different micro-climatic conditions (see Supplementary Information6 , Fig. S6.1 and S6.2 ). Data from temperature loggers were combined with those collected by the weather station and used to build interpolation models predicting hourly and maximum daily temperature for each of the 10 × 10 m pixels in the study area at any given day within the study period (see Supplementary Information 6 for full details).
To estimate vegetation quality and quantity we used the Normalized Difference Vegetation Index (NDVI), which has been widely used to depict forage productivity in mountain ungulates24 (link),51 (link),64 (link)–66 (link), and proved to strongly correlate with faecal crude proteins in ibex23 (link). NDVI was acquired by the Moderate-resolution Imaging Spectroradiometer (MODIS) on board of the AQUA satellite (16-day-composites from daily data recorded at a 250 × 250 m pixel size).
A 10 × 10 m Digital Elevation Model (DEM, Regione Valle d’Aosta official data) was used to generate same-resolution raster files for terrain aspect (cosine-transformed to range from −1 to 1), terrain slope (in degrees), and terrain ruggedness (in meters, calculated sensu Riley et al.67 ).
A 4-level categorical land cover map based on aerial image interpretation and validated by ground surveys was provided by the GPNP (GPNP, official data). Levels were defined as follows: meadows and grassland, woods and bushes (i.e., larch and Swiss stone pinewoods, pioneer woods, invasive bush, bushes), screes and rocks, and other (i.e., abandoned crop fields, urban areas/infrastructure).
Based on previous studies on the anti-predator behaviour of mountain ungulates68 (link)–70 (link), we defined safe areas as rock and scree sites with a slope steeper than 45°. We thus calculated a raster with the distance from >45° steep safe areas as a proxy for predation risk. We repeated the same procedure with a different threshold (distance from areas with >30° slope) because, to the best of our knowledge of ibex ecology, our definition of safe areas slightly differed from the one currently accepted. Finally, we created a raster of the distance from the closest hiking trail as a proxy for human disturbance, along with the estimate of the average number of hikers using that trail (GPNP, official data). Hiking trails have been historically outlined to avoid rugged terrains, to maximize wildlife sighting, and typically lay at bottom of U-shaped valleys. Therefore, hikers walking on the trails are easily detected by ibex present in the valley, from here the clear association of trails to human presence by ibex71 (link).
Fine-scale temperatures were recorded hourly in the Levionaz Valley using temperature loggers (iButton DS1922L, Maxim Integrated, n = 15 in 2010, n = 17 in 2011) stratified by elevation and hydro-geographic sectors corresponding to different micro-climatic conditions (see Supplementary Information
To estimate vegetation quality and quantity we used the Normalized Difference Vegetation Index (NDVI), which has been widely used to depict forage productivity in mountain ungulates24 (link),51 (link),64 (link)–66 (link), and proved to strongly correlate with faecal crude proteins in ibex23 (link). NDVI was acquired by the Moderate-resolution Imaging Spectroradiometer (MODIS) on board of the AQUA satellite (16-day-composites from daily data recorded at a 250 × 250 m pixel size).
A 10 × 10 m Digital Elevation Model (DEM, Regione Valle d’Aosta official data) was used to generate same-resolution raster files for terrain aspect (cosine-transformed to range from −1 to 1), terrain slope (in degrees), and terrain ruggedness (in meters, calculated sensu Riley et al.67 ).
A 4-level categorical land cover map based on aerial image interpretation and validated by ground surveys was provided by the GPNP (GPNP, official data). Levels were defined as follows: meadows and grassland, woods and bushes (i.e., larch and Swiss stone pinewoods, pioneer woods, invasive bush, bushes), screes and rocks, and other (i.e., abandoned crop fields, urban areas/infrastructure).
Based on previous studies on the anti-predator behaviour of mountain ungulates68 (link)–70 (link), we defined safe areas as rock and scree sites with a slope steeper than 45°. We thus calculated a raster with the distance from >45° steep safe areas as a proxy for predation risk. We repeated the same procedure with a different threshold (distance from areas with >30° slope) because, to the best of our knowledge of ibex ecology, our definition of safe areas slightly differed from the one currently accepted. Finally, we created a raster of the distance from the closest hiking trail as a proxy for human disturbance, along with the estimate of the average number of hikers using that trail (GPNP, official data). Hiking trails have been historically outlined to avoid rugged terrains, to maximize wildlife sighting, and typically lay at bottom of U-shaped valleys. Therefore, hikers walking on the trails are easily detected by ibex present in the valley, from here the clear association of trails to human presence by ibex71 (link).
Free full text: Click here
