Selection for seed yield in lolium perenne L.

The results of one cycle of phenotypic recurrent selection within a population of Lolium perenne showed that: (1) Selection for high seed yield led from 7.14 g/plant (unselected control) to 14.67 g/plant with a realized heritability of 40%; (2) Selection for 1000-seed weight gave an absolute response of 156.4 mg (10.1% of the unselected control) with a realized heritability of 43%; (3) Response to selection for early and late date of ear emergence was markedly asymmetrical, the realized heritabilities being equal to 89.1% and 33.0% respectively; (4) The analysis of correlated responses revealed that selection for seed yield was accompanied by a shift towards earliness of 3.6 days and by a 7.2% increase of 1000-seed weight; selection for 1000-seed weight increased earliness by 2.0 days and seed yield by 54.5%. Asymmetrical correlated responses were observed with selection for early and late date of ear emergence. Selection for earliness did increase seed yield (112.7%) and did not affect 1000-seed weight while selection for lateness decreased 1000-seed weight (9.2%) without affecting seed yield. It is concluded that the best selection criterion to increase seed yield and to avoid adverse effects upon forage production should be 1000-seed weight. A breeding strategy involving alternate cycles of selection and evaluation of forage yield is envisaged.