Spatiotemporal patterns of resource selection by free-roaming (feral) horses

Authors: PETERSEN, STEVEN - Brigham Young University; Clark, Patrick; COLLINS, GAIL - Retired Non ARS Employee; GOMEZ, TERESA - Brigham Young University; CARR, CRAIG - Montana State University

Submitted to: Rangeland Ecology and Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/20/2026
Publication Date: N/A
Citation: N/A

Interpretive Summary: The unique relationship between humans and free-roaming feral horses creates a socio-ecologic mismatch which heightens the importance of better understanding horse habitat selection patterns and areas at risk to adverse impact. We studied horse habitat use by delineating herd home range boundaries, modeling individual herd and population resource selection, and validating the predictive performance of overall, seasonal and annual resource selection models. We found that horses consistently select areas in close proximity to either stream or springs and wet meadows throughout the year and avoid mountain big sagebrush communities. Our research will aid managers of wildlife refuges, herd management areas, and other landscapes where feral horses roam and require specialized and informed management and policy support.

Technical Abstract: Free-roaming horses are revered by humans as icons of culture, heritage, labor needs, and recreational opportunities throughout western landscapes. They have also been shown to compact soils, alter plant communities, increase vulnerability to weed invasion, and reduce populations of competing wildlife. This unique relationship between humans and horses has created a social-ecological mismatch that influences equine population patterns, habitat conditions, space use, and efforts to improve western North American rangelands. We studied horse habitat use by delineating herd home range boundaries, modeling individual herd and population resource selection, and validating the predictive performance of overall, seasonal and annual resource selection models. Between 2009 and 2012, we captured and collared 9 lead mares with GPS collars at the Sheldon National Wildlife Refuge in Northwestern Nevada. Dynamic Brownian bridge models were used to delineate horse home ranges and any migratory extents. We investigated horse resource selection within these home ranges by developing 23 spatially explicit environmental co-variates Resource Selection Function (RSF) model construction. Models were fitted using binomial regression. Seasonal and global performance of models highly ranked by AICC scores, were evaluated using reserved validation data and Spearman correlation scoring. The average home range per collared horse during sample period was 42.3 ± 8.1 km2. Home ranges of all 9 herds overlapped to some degree and six herds expressed over 90% overlap with the composited home ranges of the other herds. The top models for predicting horse resource use included the following variables: distances to streams and springs, proximity to roads, elevation, and use of wet meadow and mountain big sagebrush habitats. We found that horses consistently select areas in close proximity to either stream or springs and wet meadows throughout the year and avoid mountain big sagebrush communities. The selection of stream and spring systems may exacerbate ecological impacts to water sources as well as impact wildlife in these areas through direct competition. While some horses avoided roads, other preferred areas near roads, likely because of increased travel efficiency. Compared to other vegetation types, horses consistently selected wet meadow (riparian) habitats. Wet meadow plant communities support high biodiversity compared to the surrounding landscape including forage plants such as perennial grasses and forbs. They also avoided mountain big sagebrush plant communities, likely because of the difficulty of traveling through dense tall shrubby areas when more suitable habitats with shorter plant structure is present across the landscape.