Canonical Correlation Analysis of Interactions in Insect-Infested Stored Wheat 1

Abstract

Multivariate data obtained at triweekly intervals for 60 wk from 8 insect-infested wheat bulks, 157 kg each maintained at 30°±2°C, was interpreted with canonical correlation analysis. One system consisting of 2 bulks of insect-free wheat served as a control; a 2nd system of 3 bulks was artificially infested with Rhyzopertha dominica (F.), Sitophilus oryzae (L.), and Tribolium castaneum (Herbst) (RST); and a 3rd system of 3 bulks was infested with Cryptolestes ferrugineus (Stephens), Oryzaephilus surinamensis (L.), and Tribolium castaneum (COT). Low populations of a mite, Tarsonemus granarius Lindquist were naturally present in all bulks. Canonical correlation analysis of abiotic and biotic variables identified and quantified interrelations within the systems. The canonical correlation of the first set of canonical variables ranged from 0.906–0.936 in the 3 systems. The 1st 3–4 pairs of canonical variables were significant ( P <0.01) in all of the systems; for the control system, the 1st pair of canonical variables revealed that the fungus Alternaria alternata (Fr.) Keissler and bacteria were closely related to fat acidity values (FAV) of the wheat, seed germination, dust weight and O2 levels. In the RST system, Alternaria , bacteria, and Sitophilus were related to FAV, CO2 levels, grain moisture, germ and endosperm damage, and seed germination. In the COT system Alternaria , bacteria, Tribolium and Cryptolestes were related to the variables FAV, CO2, dust weight, germ damage and seed germination.