Under global climate change and intensified human activities, habitats in northwestern China's Altay region-comprising forests, grasslands, deserts, and alpine areas-are increasingly fragmented. This study assessed the suitable habitat distribution and fragmentation characteristics of three ungulate species: red deer, goitered gazelle, and Siberian ibex. An ensemble species distribution model based on BIOMOD2 was used, and fragmentation metrics were calculated with the landscapemetrics R package. Red deer had the largest suitable area (20,688.54 km & sup2;), followed by goitered gazelle (16,342.25 km & sup2;) and Siberian ibex (11,477.62 km & sup2;). Fragmentation analysis revealed that red deer habitat patches in forest areas had the highest patch density (PD = 0.042); grasslands showed the largest dominant patches (LPI = 63.31) and highest aggregation (AI = 74.45). Gazelle habitats in deserts were the least fragmented (PD = 0.017, AI = 92.06), while ibex habitats in grasslands had higher fragmentation (PD = 0.046) but moderate aggregation (AI = 75.31). The results reveal clear interspecific differences in habitat suitability and landscape patterns, highlighting species-specific responses to environmental gradients. This study provides a scientific basis for habitat management, conservation planning, and connectivity enhancement for large herbivores in cold-arid ecotones.