About Us | Help Videos | Contact Us | Subscriptions



This article in CS

  1. Vol. 35 No. 6, p. 1536-1541
    Received: Feb 3, 1995

    * Corresponding author(s): csneller@comp.uark.edu
Request Permissions


Comparing Soybean Cultivar Ranking and Selection for Yield with AMMI and Full-Data Performance Estimates

  1. Clay H. Sneller  and
  2. Don Dombek
  1. Dep. of Agronomy, Univ. of Arkansas, Fayetteville, AR 27201



The additive main effects and multiplicative interaction (AMMI) estimate of genotype performance removes genotype × environment interaction (GEI) noise that is intrinsic to full-data estimates of genotype performance in specific environments, thereby producing estimates that may be more predictive of future performance than full-data estimates. Our objectives were to (i) compare genotype rankings based on AMMI and full-data estimates of performance, and (ii) compare the yield of genotypes selected by the two sets of estimates. Five data sets, each containing data from two consecutive years, were constructed from yield trials of Maturity Group V soybean [Glycine max (L.) Merr.] cultivars conducted at five locations. The AMMI and full-data estimates of genotype performance were calculated for each location and year in each data set. Rank correlations were calculated between the AMMI and full-data estimates. The yield of the top 15% of all genotypes and the top genotype, as ranked by each method in each location and each year, were compared in the other year of the data set. All correlations between AMMI and full-data estimates were significant within each environment, and both estimates appeared equally predictive of future genotype rankings. The yields of genotypes selected based on either set of estimates were essentially equivalent in nonmodeling years. There was no advantage or disadvantage to either method of estimating genotype performance. Apparently, the GEI patterns incorporated in the AMMI estimates in our study were not repeatable enough across years to produce estimates that were more predictive of future performance than full-data estimates.

  Please view the pdf by using the Full Text (PDF) link under 'View' to the left.

Copyright © .