Use of a Latin square design to assess experimental errors in field evaluation of cotton for resistance to Fusarium wilt Race 4

Abstract

AbstractField evaluation of cotton for resistance to Fusarium oxysporum f. sp. vasinfectum Race 4 (FOV4) is important in breeding. In this study, an 8 × 8 Latin square design was arranged in two FOV4‐infested fields each with two trials to evaluate eight cotton (Gossypium hirstum) cultivars based on mortality rate (MR) and root vascular staining (RVS). For a comparison, two trials of the same design were grown in the greenhouse. Consistent with the mapping pattern of MR in the Texas field, the analysis of variance confirmed the existence of effects due to tiers not planting rows in the Texas location only. Significant variation due to cultivars was consistently detected in the two trials of both locations. The repeatability was confirmed by significant correlation coefficients between most tiers (replications) and between the two trials in each field. The greenhouse test had the highest mean MR (72–78%) and lowest coefficients of variation (CV; 17–28%), followed by the Texas field with a lower MR (36–57%) and similar CV (22–29%); and the New Mexico test had the highest CV (59–73%) and the lowest MR (13–18%). The MR between the two fields and between the fields and the greenhouse were significantly correlated. ‘Pima S‐7’ had the highest MR (79–93%), whereas ‘FM 2334GLT’ had the lowest (15–35%), consistent with RVS at crop maturity. Among randomized complete block designs (RCBD) with three to seven replications, the mean squared error and CV were the lowest in RCBD with three replications.