Comparison of Statistical Methods to Assess Yield Stability in Sweetpotato

Abstract

Twenty-seven trials were conducted with 20 sweetpotato [Ipomoea batatas (L.) Lam.] clones in seven locations over 4 years in Cameroon to provide data for the evaluation of statistical methods appropriate for assessing yield stability. Four stability assessment methods were used. Clones 048, TIb 1, 1639, TIS 2498, and TIS 2544 were identified as stable for storage root yield by all four methods. The Eberhart and Russell and Perkins and Jinks methods are statistically related, each detected 9, 8, 9, and 10 clones as stable for total yield, marketable yield, and total and marketable root counts, respectively. Some clones considered stable for yield were found to be unstable for root counts. This result suggests that stability of a complex trait (such as yield) may not depend on stability of component traits. The Shukla method and the Francis and Kannenberg method were considered to be the most effective and useful of those tested. The Shukla method found more clones to be stable; also, the very high-yielding and very low-yielding clones were judged to be unstable. The Francis and Kannenburg method was very convenient, easy to use, and was appropriate for grouping clones of greater and lesser desirability and stability.