Phenotypic Recurrent Selection for Multiple‐Parted Flower Heads in Red Clover¹

The objective of this investigation was to ascertain the variation in and the effectiveness of phenotypic recurrent selection for multiple‐parted heads in red clover ( Trifolium pratense L.). Additional objectives were to examine the effect of changes in this character on seed yield and other morphological and physiological characters. Six generations of selection were conducted on an annual basis under field conditions involving 1040 to 2548 plants and selection intensity of 2.0 to 6.8%. The effectiveness of selection was evaluated using spaced plants established from remnants eeds. Number of head parts in largest heads increased linearly from 1 to 7.4 with a regression (b) coefficient of 0.92 head parts/generation. The check cultivar Kenstar from which the original selections were made, had 93% single‐parted heads whereas Generation 6 had only 37% single‐parted heads and 9% of the heads with 5 or more parts. Number of seeds per plant, persistence, and vigor (forage yield) declined over generations, probably due, in part, to inbreeding depression and partly due to other factors. Extreme selections for multiple‐parted heads exhibited weak and lodged, fasciated stems containing many growing points. Recurrent selection was quite effective, but changes associated with increasing multiple‐parted heads were deleterious and mayb e difficult to overcomeS. election is a powerful tool that can generate genotypes not previously seen, and conceivably could reduce the need of plant exploration for desirable genes, at least in cross‐pollinated species.