Objective  The aim was to explore the soil particle distribution characteristics and soil nutrient properties in different forest stands in the red soil hilly area, in order to provide a theoretical basis for forest stand upgrading, soil fertility improvement and sustainable management of pure forests in the region.

  Method  Three typical forest understory soils in red soil hilly areas were used as research objects, and correlation analysis and stepwise regression analysis were applied to explore the particle composition, fractal dimension and their interrelationships with soil nutrients in different forest types.

  Result  ① The soil textures of the three forest stands in this study area were mainly concentrated in silt loam, silt and sandy loam. The fractal dimension of the three stands was: eucalyptus × red cone mixed forest > eucalyptus pure forest > red cone pure forest. ② Except for total soil phosphorus, the different stand types and soil layers had significant effects (P < 0.05) on soil nutrient content, especially in mixed forests. However, the reciprocal action between stand types and soil layers only had significant effects (P < 0.05) on organic matter, total nitrogen and total potassium, and had no significant effects on pH and total phosphorus. ③ Soil particle fractal dimension was highly significantly positively correlated with clay particle content and negatively correlated with soil meal, organic matter and total nitrogen content (P < 0.05), while soil particle fractal dimension was highly significantly linearly correlated with clay, meal, very fine sand, total nitrogen and total potassium (P < 0.001).

  Conclusion  The fractal characteristics of soil particles could reflect the distribution characteristics of soil particles and their nutrient contents in different forest types, and the soil fractal dimension and nutrient contents of mixed forests were higher than those of pure forests. So mixed forests had better water, soil retention functions and higher soil fertility quality levels than pure forests.