Abstract: By using fractal model, this paper studied the fractal dimension of soil aggregate structure (D) in the slope farmland (CK), its 5-year de-farmed Neosinocalamus affinis plantation (NAP), Bambusa pervariabilis×Dendrocalamopsis oldhami plantation (BDP), Alnus crenastogyne +Neosinocalamus affinis  plantation (ANP), and abandoned farmland (AFL) in south Sichuan Province of China, and analyzed the relationships between the Dand soil physical and chemical properties. In the de-farmed plantations and abandoned farmland, the contents of ＞0.25 mm soil aggregates and waterstable aggregates were increased significantly, compared with those in the slope farmland. The D was 1.377-2.826, being in the order of NAP＜BDP＜ANP＜AFL＜CK, and decreased with the increasing contents of ＞0.25 mm soil aggregates and water-stable aggregates. Comparing with CK, de-farming increased the soil natural water content, capillary porosity, and contents of soil organic matter, total N, alkali-hydrolysable N, total P, and total K, and decreased soil bulk density, non-capillary porosity, and aeration porosity. There were close relationships between the fractal dimension of soil aggregate structure and the soil physical and chemical properties. All the results suggested that the de-farming of slope farmland was beneficial to the increase of the contents of ＞0.25 mm soil aggregates and water-stable aggregates, and the enhancement of soil structure stability. The D could be used as an ideal index to evaluate soil fertility, and planting Neosinocalamus affinis  on the de-farming slope farmland was a good measure for the improvement of soil fertility in the research area.

Key words: de-farming pattern, vegetation restoration, aggregate structure, fractal dimension, Baotianman warm temperate forest, soil bacterial community, 16S rRNA, Illumina Miseq.