About
  PDF
Full Text
(78 K)

Use of Regression Analysis in Variety Selection

David W. Albers, Tom Kerby and Constance Garcia


ABSTRACT

Variety comparisons by growers are critical decision for each production year. Summarizing by averaging together tests across a geographic area will determine which variety had the highest yield over all the locations tested, but may not determine the best variety to plant on any particular farm or field environment. The procedure outlined here allows relative variety performance to be measured across a range of yield environments. Use of the D&PL Performance Database as a data source, followed by linear regressions of lint yield and crop values of “Variety 1” on “Variety 2”, was able to determine what portion of the yield spectrum and crop value spectrum each variety is favored. In the examples used, DP 451 BR is favored over ST 4892 at yield levels up to 746 lb/acre and crop values up to $571 / acre. PM1218 BG/RR was favored over ST 4892 BR at yields up to 1337 lb/acre and crop values up to $817 / acre. PM 1218 BG/RR was also favored over DP 451 B/RR over the normal range of cotton yields, and up to $871 / acre crop value. This approach does not remove the need to assess varieties for adaptation to local conditions such as soils, diseases, and other factors, but does allow a broader view of variety performance that typical multi-location and multiyear summaries do not permit.





[Main TOC] | [TOC] | [TOC by Section] | [Search] | [Help]
Previous Page [Previous] [Next] Next Page

Document last modified May 20, 2002