Agronomic performance of rice breeding lines selected based on plant traits or grain yield

Selection for yield per se has greatly contributed to yield improvement in many crops. It is expected that selection based on plant traits is more effective in increasing crop yield potential. This study was conducted to compare the effectiveness of trait-based and yield-based selection in increasing rice yield and to determine whether lines with ideotype traits have the potential to express higher yield under optimal crop management conditions. Lines were selected based on plant traits or on grain yield measured in a breeder's replicated yield trial. The main target traits for selection were plant height, leaf and panicle morphology, grain size, total dry weight, and grain-filling percentage. Yield performance of trait-based selection was compared with that of yield-based selection in an agronomic trial with optimum crop management for three seasons. Trait-based selection increased leaf area index and total dry weight but reduced spikelet number per m(2) and harvest index compared with yield-based selection. Consequently, selection based on plant traits did not increase grain yield compared with selection based on yield per se. In one of the three seasons, yield of trait-based selection was significantly lower than that of yield-based selection. Among all tested breeding lines, maximum yield was produced by yield-based selection and minimum yield came from trait-based selection. These results suggest that lines with ideotype traits did not express higher grain yield than lines selected based on yield per se under optimal crop management conditions, and yield-based selection was as effective in increasing rice grain yield as trait-based selection in the late generations of the breeding cycle. (C) 2011 Published by Elsevier B.V.