CHOOSING SOYBEAN VARIETIES FROM YIELD TRIALS - MULTIPLE MATURITY GROUPS AND YIELD VARIABILITY

Soybean [Glycine max (L.) Merr.] variety trials are used to compare re lative yields among soybean varieties. The value of these trials deriv es from their ability to predict which varieties producers should grow in the next year. The objective of this study was to evaluate the pre dictability of 20 different variety selection methods, not only for ma ximum yield but also for the variability in yield. Eighteen selection methods involved constantly changing varieties while two methods (stab le methods) reduced the number of replacement varieties. Data from the Kentucky Soybean Performance Tests 1988-1994 were analyzed. The yield increases (mean of 28 environments) produced by the first 18 methods ranged from 4.0 bu/acre (8%) to 1.1 bu/acre (2%); 13 of the 18 methods produced increases significantly greater than zero. Averaged across a ll other factors, the selection of one variety provided significantly higher yield increases (+2.9 bu/acre) than the selection of three vari eties (+1.9 bu/acre), but the yield variation was significantly greate r when one variety was selected than when three varieties were selecte d, particularly when the three varieties were selected from three diff erent maturity groups. The stable methods involved growing a selected variety until its yield fell below the statewide I-yr mean yield of al l varieties in the test. The stable methods had a variety turnover rat e of approximately 25%, compared with approximately 80% for the other methods. These stable methods provided yield increases (+2.3 bu/acre f or selecting one variety and +1.8 bu/acre when three varieties were se lected) and yield variability equal to the other methods. The stable m ethod of selecting three varieties, one from each maturity group, appe ars to provide a compromise between yield benefits and yield variabili ty while reducing variety changes.