Abstract
This study investigates the spatial heterogeneity and species diversity of grassland vegetation in the agro-pastoral ecotone of China's Loess Plateau, integrating Taylor's power law model with the minimum area concept to address scale-dependent ecological patterns. Field surveys were conducted across four vegetation types: small-leaf poplar forest (SP), pine-caragana mixed forest (PC), caragana shrubland (RC), and saline grassland (SG). Nested quadrats (0.25-8 m(2)) were used to establish species-area relationships (SARs), while binary occurrence frequency data fitted to Taylor's power law quantified spatial heterogeneity parameters (δ(i), δ(c), CACD) and derived diversity indices (H', J', D). the results showed that species composition differed significantly among vegetation types, with RC exhibiting the highest richness (25 species) and SG the lowest (12 species). SAR analysis showed distinct z-values: SP had the lowest z (0.14), indicating minimal area effects and high homogeneity, while SG had the highest area sensitivity. Spatial heterogeneity (δ(c)) was highest in RC and lowest in SP. Over 82.5% of herb-layer species exhibited aggregated distributions (δ(i) > 0). The dominant species Leymus secalinus (Georgi) Tzvelev shifted from regular (δ(i) < 0) under SP/SG to aggregated (δ(i) > 0) under PC/RC. Diversity metrics peaked in PC plots (highest H' and richness, lowest dominance), whereas SP showed high dominance but low diversity. CACD values (critical aggregation diversity) were maximized under SG. The integration of power law modeling and minimum area analysis effectively captures scale-dependent vegetation patterns. Pine-caragana mixed forests (PC) optimize biodiversity and spatial heterogeneity, suggesting moderated canopy structures enhance ecological stability. These findings provide a theoretical basis for sustainable grassland management in ecologically sensitive agro-pastoral zones.