STRUCTURED HETEROGENEITY OF SOIL-PH AND GRAIN-YIELD OF RICE AND WHEAT GROWN IN A SODIC SOIL

Conventional statistical methods that assume independent random variation are inadequate when dealing with spatially dependent heterogeneity. Structural analysis of heterogeneity was performed for soil pH and grain yields of rice (Oryza sativa L.) and wheat (Triticum aestivum L.) grown under uniform management in a sodic soil (Natric Haplustalf). Relative efficiencies of field sampling designs and methods of analysis were computed. The coefficients of variation (CV) for the yield of rice and a subsequent wheat crop were 16 and 54%, respectively. The two-way trend variation in soil pH was 21% of variance before rice planting and 28% after wheat harvest. The same parameter for the grain yield of rice and wheat was 12 and 28%, respectively. Variogram analysis revealed that locally structured variation of soil pH was 30 and 48% of variance before rice planting and after wheat harvest, respectively. Structured heterogeneity for grain yield was 16% in rice and 59% in wheat. Soil pH heterogeneity effects on rice were of lower magnitude than with wheat. The relative efficiency of Wilkinson's nearest neighbor analysis as compared to a randomized block design varied from 623 to 1682% depending on size and shape of the plot. This indicates that a substantial improvement in experimental precision is possible by taking account of positional effects of observation plots.