Selection indices for improving selection efficiency in Indian mustard

In Indian mustard (Brassica juncea L. Czern & Coss), a major oilseed brassica crop, oil yield is the ultimate trait of economic value, which is determined by seed yield (SY) and oil content (OC). SY, the end product of many complex biological processes, is influenced by many traits, such as plant height (PH), branch number, siliqua number, seed weight, seed size and pod length. It is difficult to accurately select for high-yielding genotypes based exclusively on SY, neglecting other yield-determining traits in commercial breeding programs. Therefore, it is desirable for breeders to simultaneously improve several traits of economic importance, most of which are quantitative in nature. The purpose of the present study was to develop selection indices for simultaneous selection of multiple quantitative traits to enhance selection efficiency of the mustard breeding program of Punjab Agricultural University (PAU) located in Abohar, Punjab. A set of 487 diverse mustard genotypes, comprising germplasm accessions of Indian and European origin, advanced breeding lines, landraces, and cultivars, was evaluated using an alpha lattice design for 2 years (2015-16 and 2016-17) at the Regional Research Station of PAU, Abohar, District Fazilka (the major mustard-growing region of the Punjab state). Data on PH, main shoot length (MSL), number of primary branches (NPB), number of secondary branches (NSB), number of siliquae on main shoot (NSMS), siliqua length (SL), number of seeds per pod (NSP), 1000-seed weight (TSW), SY on a plot basis and OC were recorded. To compute selection indices, economic weights, as determined via path coefficient analysis, were assigned to major yield-determining traits. Out of eight recorded agronomic traits, four traits, viz., NSMS, TSW, NSB and NPB, had larger direct effects on SY than other traits, and hence were assigned relative economic weights of 0.60, 0.20, 0.12 and 0.08, respectively. Selection indices were computed using different combinations of traits via Kang_SASPath computer program. When seven traits were used in a selection index (PH, NPB, NSB, NSMS, SY, TSW and MSL; or PH, NPB, NSB, NSMS, SY, TSW and OC), expected genetic advance (GA)at 5% selection intensity was about the same (9.55; selection efficiency or SE = 163% compared with selection based on SY alone).The GA was 9.46 (SE = 161%)for five traits, viz., NPB, NSB, NSMS, SY and TSW; and 9.30 (SE = 159%)for four traits, viz., NPB, NSB, NSMS, and TSW. These GAs were compared with the GA for SY only, which was 5.90 (benchmark; SE = 100%). We concluded that simultaneous selection based on multiple traits was always more efficient than selection based on SY alone. As there was only a small difference between the GA based on four traits (SE = 161%) and GA based on seven traits (SE = 163%), we suggest that mustard breeders use the parsimonious combination of four traits, viz., NPB, NSB, NSMS, and SY, to enhance SY and, consequently, oil yield in mustard.