Grazing animals can cause nutrient transfer if spatial patterns of nutrient ingestion and nutrient return differ. This is the case if different factors drive spatial preferences for grazing and non-grazing activities. While nutrient transfer has often been reported from large-scale grazing systems, it should be less pronounced in smaller pastures, where reduced site heterogeneity and decreased distances to water lead to more uniform distributions of grazing animals. We investigated the differences between spatial patterns of grazing and non-grazing behavior and their impact on soil nutrient concentrations in a 15-year-old cattle grazing experiment. This compared three grazing intensities with three replicated 1-ha paddocks each. Long-term patch-grazing had led to a pronounced mosaic structure of short and tall vegetation and to a differentiation of vegetation composition between these patches. We recorded locations and activity of cattle in 1-min intervals using GPS telemetry combined with activity sensors. Activity sensor readings were classified by k-means clustering and validated by direct animal observation. Sward structure was predicted from high-resolution aerial photographs using random forest classification. Topsoil nutrient concentration was determined at georeferenced points in short and tall patches.
Active behavior (grazing, walking) could successfully be distinguished from inactive behavior (standing, lying down) based on activity sensor readings (validation error < 10%). Texture parameters were important for the classification of the aerial photograph into short and tall patches (validation error < 10%). Estimated densities of active behavior, but not of inactive behavior, were higher in short than in tall patches. GPS locations of inactive behavior were highly clustered in hot spots that were largely independent from topography or location of water supply. A nutrient transfer index (difference between standardized densities of active behavior and total animal locations) was positively correlated with soil nutrient concentrations, implying that substantial long-term nutrient transfer through grazing animals had occurred. Nutrient depletion of preferentially grazed short patches and nutrient accumulation at preferred resting sites have implications for productivity and botanical composition of the investigated pastures. We found patch-scale and paddock-scale differences in grazing and non-grazing behavior that corresponded to measured soil nutrient concentrations even in relatively small pastures. Nutrient transfer at ecosystem function relevant scales may therefore be a pervasive phenomenon of large herbivore grazing systems independent of their spatial scale.