Genotype � environment interactions in symptom development and yield of cassava genotypes with artificial and natural cassava bacterial blight infections

Abstract

Thirty-seven cassava genotypes from Benin, including advanced breeding lines, were tested for their reaction to bacterial blight in the forest–savanna transition, wet savanna and dry savanna zones of Benin. Sixteen genotypes were repeated in 12 environments. In year 1998, genotypes RB92164, RB92022, TMS30572, BEN86004, RB92033 and Dangbo2, and in year 2000, genotypes RB92202, RB92151, RB92132 and TMS30572 were resistant in one ecozone. Among the more resistant genotypes, CAP94030, BEN86040, RB89509, RB92132 and TMS30572 showed low interaction across environments and were most stable in disease reaction. Ten genotypes were classified as high yielding across environments. Among the more resistant group of genotypes, only TMS30572 and RB89509 were high yielding, with RB89509 being unstable in yield across environments. Selection of genotypes proved reliable only after artificial inoculation. Comparing environments, artificially inoculated treatments in the wet savanna zone and in the forest–savanna transition zone with stable high symptom severity proved most suitable for screening of genotypes, while the wet savanna zone with low natural infection in year 1998 was suitable for production of propagation material, and the site in the dry savanna zone with natural infection in year 1998 was the best environment for cassava production. The correlation between disease severity and root yield was significant only for the non-inoculated treatment in the dry savanna zone in year 2000 (R = −0.58), but not in any other environment. Among the 37 genotypes tested, several genotypes can be recommended to farmers in specific ecozones, and genotype TMS30572 revealed as relatively stable in disease resistance and in high yield across ecozones.