A farmer-friendly method of determining the most suitable cultivation soils would help in transferring new integrated soil management technologies. The potential for using soil color (Munsell data) was tested by physico-chemical analysis of 1028 ferrallitic soil samples from 289 profiles unequally allocated above 7 deg N in Côte d’Ivoire. Soil hue variations in depth and along the toposequence revealed the existence of vertical and lateral gradients of soil hue. The relative contribution of the different descriptors (clay, sand, carbon, total nitrogen, total phosphorus, potassium, magnesium and calcium) to the three functions extracted using a discriminant analysis to differentiate the four groups of soils with different hues was evaluated as well as the analysis of variance to determine the possible groups number for each one of the descriptors. Differences between physico-chemical components of red (2.5YR and 5YR) and yellow (7.5YR and 10YR) soils were determined, especially for P, Mg and K in extension. A decreasing gradient of inherent soil fertility indicators with an increasing yellowness in soil hue was revealed using multiple regression models. The soils 2.5YR and 5YR were therefore deemed more appropriate for stable and sustainable agriculture.Key words: Hue, ferralsols, fertility, soil use
Discriminant Analysis for Regression Models with Stationary Long-Memory Disturbances