This study aimed to clarify the effects of row spacing on forage yield, quality, and soil function in an oat-forage pea intercropping system in the high-altitude Qinghai-Tibet region. A two-year, multi-site field experiment was conducted in Hainan and Haixi Prefectures, in which DTOPSIS and structural equation modeling (SEM) were integrated to evaluate multiple objectives. The results indicated that the optimal row spacing differed between sites. In Hainan, a row spacing of 30 cm maximized forage yield (11,507.94 kgha(- 1)), representing a 12% increase compared with 25 cm spacing, although the soil quality index (SQI) declined by 9.67%. In Haixi, a 25 cm spacing enhanced forage yield (6,905.98 kgha(- 1)) and maintained a higher SQI than wider spacings. SEM analysis identified site-specific pathways linking row spacing, forage productivity, and soil function, and emphasized the presence of negative feedback mechanisms in nutrient-limited or arid environments. The findings suggest that optimal row spacing is region-specific, balancing trade-offs among yield, forage quality, and soil sustainability, and providing mechanistic insights for precision spatial management of alpine intercropping systems.