Premium
Selection of Concomitant Variates Affecting Regrowth, Yield, and Digestibility in Forage Sorghums 1
Author(s) -
Fribourg Henry A.,
Creel Rodney J.
Publication year - 1981
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj1981.00021962007300030014x
Subject(s) - agronomy , dry matter , forage , sorghum , sowing , cultivar , biology , mathematics , leaf area index , crop , sweet sorghum
Numerous environmental factors and plant characteristics, and interactions among them and with management practices, affect growth and composition of summer annual forage sorghums ( Sorghum bicolor (L.) Moench). Thirty‐eight plant and environment characteristics were estimated concurrently with the dry matter (DM) yield, DM regrowth rate (DM yield/day since emergence or since preceding harvest), and in vitro dry matter digestibility of three cultivars grown in the field on a fine‐loamy, siliceous, thermic Humic Hapludults and subjected to 19 different defoliation regimes in 2 different years. The selection of a subset of the 38 observed concomitant variates could be useful for planning future research. The purpose of this paper is to report on the sequential use of factor analysis and multiple regression techniques to decrease the number of independent variables without appreciable loss in the explanation for variability in the dependent variables. Factor analysis was used first to reduce possibly redundant concomitant variables or select proxy variables for a whole dimension. Fifteen of the original 38 variates were selected. multiple regression was then used to reduce further the concomitant variables to nine variables (days of growth since planting, days of regrowth since previous harvest, height of standing crop, leaf area index (LAI) of whole plant, DM yield of stubble at previous harvest, cumulative precipitation, degree days since previous harvest, percent leaf in DM harvest, LAI of whole plant at previous harvest). Six or fewer of these variates were associated with a large portion of the variability in the three dependent variables. The goal of parsimoniously identifying the variables which affected the dependent variables within a genotype in these experiments was attained. The strategy employed—sequential use of factor analysis and multiple regression for reduction in number of concomitant variables—may have merit for investigations in many agronomic disciplines.