Covariance Structure among Yield-Components and Morphological Traits for Oilseed Genotypes



Factor-Analysis was conducted using estimation methods: Principal-Components-Factor-Analysis (PCFA) and Maximum- Likelihood-Factor-Analysis (MLFA) to investigate the covariance-structure which may exists among physical/morphological and yield component traits and to replace the system of traits into two or three factors that can represent and explain maximum of variation present in original system of traits for Brassica-Napus, Helianthus-Annuus and Arachis-Hypogaea. Factor-Analysis on multi-trait of Brassica-Napus data showed that first four factors extracted 85.45% and 72.3% of total variation present in the original system of traits using PCFA and MLFA respectively. Both methods produced approximately same loadings for traits for factor-1 and factor-4, while factor-2 and factor-3 were observed slightly different. Traits such as: days-to-maturity, plant-height, branches/plant and silique/plant all have high positive loadings for factor-1 in both methods and so this factor might be labeled as biological-yield whereas factor-2 appeared as contrast in days-to-flowering with seeds-weight in PCFA but days-to-flowering has only dominating loading for factor-2 using MLFA so it can be labeled as flowering-time factor. Similarly, factor-4 turned out as oil-yield factor in both methods. The estimated residual matrices (R 􀵆 L􀷨L 􀷨 􀵆 Ψ􀷩 􁈻, showed that MLFA estimates of loadings􁈺L􀷠) and specificvariances 􁈺Ѱ􀷡 􀭧􁈻 did better job in reproducing R as compared to PCFA estimates. So comparison based on residual-matrices, possible factors extracted and interpreted using MLFA can be considered more reliable. In context of Helianthus-Annuus multi-trait data, both methods showed that first four factors extracted 87.4% and 73.5% of total variability respectively and produced same loading pattern for first three factors while loading pattern for fourth factor was turned out different. Days-to-flowering-initiation, days-to-floweringcompletion and days-to-maturity all have high positive loadings for factor-1 in both methods so it might be labeled as floweringduration factor whereas in second factor: head-diameter and %oil-contents both have high loading and it might be regarded as yield factor in both methods. In comparison of methods w.r.t. specific-variances and estimated residual-matrices, PCFA was found to be more appropriate as compared to MLFA. Similarly, factor analysis using Arachis-Hypogaea data revealed that PCFA extracted three factors and captured 83.4% while MLFA extracted four factors with 80.3% of total variation in multi- trait system. Moreover, shelling% and SMK% showed high positive loadings for factor-1 in both methods so it might be labeled as Arachis-Hypogaea-yield factor. Also in factor-2, germination% and days-to-maturity, both showed high loading in PCFA and only days-to-maturity showed high loading in factor-2 using MLFA and it might be regarded possibly as time-to-biological-yield factor in both estimation methods.

Full Text:



  • There are currently no refbacks.

Copyright (c) 2016 Sindh University Research Journal - SURJ (Science Series)

 Copyright © University of Sindh, Jamshoro. 2017 All Rights Reserved.
Printing and Publication by: Sindh University Press.