Spatial Heterogeneity Of Soil Properties In Secondary Broadleaved Forest

Spatial heterogeneity is one of the soil natural characters. Soil property distinguished in different spatial positions at the same time, even in plot scale from the same land use. So, It is meaning for study the characters and influential factors of forest soil to clarify the dynamic relationship between plant and soil, to improve the precision of forest soil sampling and observation in theory and in practice.We adopt traditional Statistics and Geostatistics to study soil spatial heterogeneity and influence factors in secondary broadleaved forest under natural regeneration (NR) and artificial forcing natural regeneration (AFNR.). Samples were taken in the surface soil 0—20 cm in stand scale(40 m×40 m)and plant individual scale(2 m×2 m), and 6 soil properties(pH, organic matter, total nitrogen, hydrolysable nitrogen, available phosphor, available potassium) were analyzed. The research indicated that, in stand scale under NR, stochastic proportion of spatial variation of most soil properties, result from higher density, more species, cross distribution of roots, different qualities and quantities of plant litter in different spatial positions. These variations are less depending on the scale, and have a more complicate and interlaced spatial pattern. In stand scale under AFNR, the variations are depending on the scale, different soil properties put up similar spatial pattern. These characters lie in the low density and diversity of plants in stand; target species showed good growth traits, which make nutrients enriched around them. So that's why spatial patterns of different soil properties would overlap. In plant individual scale, most soil properties display no sill figure in semivariogram, which means the ranges of most soil properties in semivariogram are larger than the longest sampling distance. Correlations and autocorrelations of most of soil properties are lower in this scale than those in stand scale under NR, whereas, the ratio of C to Co+C decreased with the scale refined.Result of forest soil sampling indicates that: design-based approach is mainly based on the traditional investigation, the number of samples and study area are mostly considered by this approach. The amount of soil samples are increase with the confidence level and sample precision, and the amount of soil samples distinguish between each soil property. Ranges of in semivariogram are mainly used in the fix on the most minimal border and area of study plot. Model-based approach is based on the Geostatistics, the sampling distances which is longer than ranges are of no effect on any interpolation and mapping.