A COMPARISON OF REPLICATED S(1) LINE PER SE AND RECIPROCAL FULL-SIB INDEX SELECTION IN MAIZE (ZEA MAYS L.)

An evaluation study was done to compare the effectiveness of ${\rm S}\sb1$ line per se $({\rm S}\sb1)$ and reciprocal full-sib selection (RFS) for improvement of a selection index and its component traits using three maize populations: Nebraska B Synthetic (B), Nebraska Krug open-pollinated variety (K), and the Nebraska Stiff Stalk Synthetic (S). ${\rm S}\sb1$ selection was done in each population while RFS selection was done between the B and K and between the B and S populations. The selection criterion in each selection procedure was the multiplicative selection index composed of yield (adjusted to 15.5% moisture content) x proportion of upright plants x proportion of undropped ears. To detect the effects of random genetic drift the selection process in each procedure was replicated three times. For the evaluation study, cycle 0 plus replicate subpopulations from cycles 3 and 5 from each of the three populations improved by either method were tested as bulked selfs, random mated and crossed populations. The evaluation studies were done under both dryland and irrigated environments over two seasons, 1985 and 1986, at Lincoln, NE. When averaged over all populations, index values of the ${\rm S}\sb1$ selected, bulked selfed populations were significantly increased by 5.8% per cycle, with a correlated increase in yield of 4.3% per cycle. RFS selection significantly increased index values in these inbred populations by 3.4% per cycle with a correlated response in yield of 1.8% per cycle. Both methods were equally effective in improving index values of the random mated populations. Inbreeding depression was reduced by ${\rm S}\sb1$ selection because the inbred response was greater than the random mated response. RFS selection was somewhat superior to ${\rm S}\sb1$ selection in improving index and yield of the population crosses. The rates of response for the index in the population crosses were 3.6% and 2.9% per cycle for RFS and ${\rm S}\sb1$ selection, respectively, when averaged over populations. Midparent heterosis was increased by both methods, but to a slightly greater extent by RFS selection. Random genetic drift was observed to cause significant differences among some of the replicate subpopulations for the index and its component traits. After five cycles of selection, the major advantage of ${\rm S}\sb1$ selection over RFS selection is its reduction of inbreeding depression. However, in breeding programs where hybrid production is the ultimate product, RFS selection might still be a better method in the long run if the variety cross can be improved by this method to a higher ultimate selection limit