Estimating yields of tropical maize genotypes from non-destructive, on-farm plant morphological measurements

Maize is the main grain crop grown in the highlands of sub-Saharan Africa, on a broad range of soil fertility and management conditions. Important yield variability has been reported at different scales, reflecting the intensity and spatial distribution of growth-limiting and growth-reducing factors. Maize yield estimation represents a valuable tool to assess within-farm variability in soil fertility through crop performance. The objective of this study was to develop mathematical relationships between plant morphological attributes and grain yield of tropical maize genotypes, based on plant allometric characteristics. These models were used to estimate maize yields and the estimates were validated against independent data collected from experimental and farmers' fields in western Kenya. Three commercial hybrids and three local varieties were considered. Multiple linear regression models including plant height and either ear length or ear diameter as explanatory variables, and simple linear regressions including only plant height, were the most accurate to estimate both total aboveground biomass and grain dry matter yields per plant (r2: 0.76-0.91). Average values for the harvest index ranged between 0.34 and 0.42, varying with the total aboveground biomass produced per plant. Yield estimations on ground area basis for farmers' fields were acceptably accurate. Plant height measurements can be easily taken at any moment after maize flowering and used in simple models to estimate maize yield. This approach proved also a valuable tool to discuss yield variability with farmers.