To understand the ecological effects of large herbivores, it is important to understand foraging behaviour and how this is affected by density. In this experiment, domestic sheep (Ovis aries) were kept at high (80 sheep per km2) and low (25 sheep per km2)population density in a landscape-scale experiment during summer in high mountain pastures in Hol, Norway. We predicted an increasing use of less preferred plant species or habitat types with increasing sheep population density. Foraging behaviour was investigated by direct observations of individually marked sheep on different spatial scales, and diet composition was also assessed with microhistological analysis of faecal samples from known individuals.
We found that effects of density on foraging behaviour were scale-dependent, and only detected at the finest spatial scale,i.e. in diet choice. Use of the common grass species Deschampsia flexuosa, which provided the bulk forage (10-65% of the diet), remained constant throughout the season at low densities, but increased significantly over time at high densities. Ewes ate a higher proportion of D. flexuosa than lambs. On a coarser spatial scale, neither patch (within vegetation type) nor habitat (vegetation type) selection was affected by density, but habitat selection differed depending on whether the sheep were grazing orresting. Our study thus provides evidence showing that density dependence in foraging behaviour occurs, but only at the finest spatial scale (diet choice). This result suggests a non-linear effect on the D. flexuosa population with an increase in sheep density: not only is an increasing number of sheep eating more, but each sheep also eat a larger proportion of D. flexuosa than at low densities. This thus provides a case showing that the ecological impact of grazers on plant populations may not be linearly related toherbivore density.