ANALYSIS OF FACTORS RELATED TO WHEAT YIELD ON FARM FIELDS IN THE BUBERUKA HIGHLANDS OF RWANDA

The wheat (Triticum aestivum L.) improvement program in Rwanda has focused on disease resistance, but other factors, including site characteristics, may impact on cultivar performance. The objectives of this study were to identify on farm fields those factors related to wheat yield that may enhance the wheat improvement strategy. Soil samples from 25 fields in 1988 and 30 in 1989, were analyzed for chemical properties. Fields were characterized by slope, azimuth, irradiance, and disease severity. Grain yield and yield components were regressed on soil and site characteristics, and disease development. Nitrogen and Al explained 35% of the variability in tiller and spike numbers for 1988; whereas, slope and stripe rust severity comprised the best model for 1989. For combined seasons, N and Al, powdery mildew and stripe rust severities, and field slope explained 55% of the variability in tiller and spike numbers. Tillers and spikes per square meter also were correlated as a second degree polynomial with irradiance and maximum numbers occurred at 28 MJ m-2 d-1. Aluminum and N were significant variables for grains per spike in 1989, and their regression coefficients usually were not significantly different among seasons. Among most yield components their standard partial regression coefficients usually were greater than for other independent variables. Azimuth and slope were model components for grains per spike, kernel weight, and grain yield. Responsiveness to N and tolerance to Al, therefore, are important selection criteria and the continued search for resistance to Puccinia striiformis West. and Erysiphe graminis DC. f. sp. tritici E. Marchal is warranted. Yields could be improved by selecting lines that tiller well under high solar flux and testing early generations on sites representative of the region.