Multivariate approach to grouping soils in small fields. I. extraction of factors causing soil variation by principal component analysis

Abstract

Surface soils were sampled from an area of about 5 ha at a density of 1 per 400 m² in the Experimental Farm of the International Institute of Tropical Agriculture in Nigeria. The samples were analyzed for physical and chemical properties including pH, content of organic carbon, available phosphorus, and exchangeable cations (Ca, Mg, and K) as well as particle size distribution (sand, silt, and clay). Principal component analysis was employed to extract the factors causing soil variation in terms of the above-mentioned properties. Four principal components accounted for 80.3% of the total variance and were named i) inherent fertility factor, ii) available phosphorus factor, iii) acidity factor, and iv) organic matter factor, respectively.

Isarithm maps based on the principal component scores of the soils for each factor reasonably reflected the surface trend of the soils under natural and cultivated conditions specific to the study area. By means of principal component analysis, variation of the soil properties was summarized into a smaller number of factors, whose scores may be used as new variables for soil grouping in a small area for research planning and experimental design.

Key Words:

• numerical classification
• principal component analysis
• soil mapping
• soil variation