Keywords: 
• soil science /
• evergreen broadleaf forest /
• soil nutrients /
• spatial variability /
• fractal dimension /
• nature reserve

Abstract: Spatial heterogeneity of soil nutrients is an important factor which affects the spatial distribution of vegetation. In this paper，geostatistical theory was used to analyze the spatial heterogeneity of soil nutrients including soil organic matter （SOM），total nitrogen （TN），available nitrogen （AN，hydrolytic nitrogen），available phosphorus （AP），and available potassium（AK） in an evergreen broadleaf forest of National Nature Reserve of Mount Tianmu，Zhejiang Province，China. Results showed that（1） an exponential model revealed the spatial structures of SOM and AN，whereas spherical models were used for TN and AP. In general， structure was the primary cause of the spatial variability，but SOM had a strong spatial autocorrelation with a spatial proportion of 0.787；TN，AN，and AP had moderate spatial autocorrelation. Due to its differences in active lag distance and its interval with different models explaining its spatial structure，AK had no clear spatial variability.（2） The maximum spatial autocorrelation for AN had a range from 4.21 to 169.50 m，whereas AP had a minimum.（3） Fractal dimensions from log-log semi-variograms quantitatively described spatial pattern differences and scale dependence of the five kinds of soil nutrients. Fractal dimensions were higher for AK，so AK spatial structure had a strong scale dependence with a complex spatial pattern. The fractal dimension of SOM was minimal， so its spatial pattern was relatively simple. Similar fractal dimensions for TN，AN，and AP explained their similar spatial patterns，but the small distinctions in their fractal dimensions did reveal local spatial structural variability.［Ch，3 fig. 3 tab. 16 ref.］