Calibration and validation of regression models for individual leaf area estimation of cauliflower grown in a hydroponic system

Abstract

Leaf area (LA) is an important parameter in many studies to evaluate plant growth, but is generally estimated by a destructive measure, i.e., requires leaves to be removed. In this context, non-destructive methods have been used based on linear measurements such as leaf length (L) and/or width (W) to determine individual LA. In this study, models were developed using linear measurements (L, W or L×W) of leaf for individual LA estimation of cauliflower grown in a hydroponic system. Two experiments were conducted, one in the autumn-winter 2019 (three cauliflower cultivars ‘Piracicaba de Verão’, ‘Sabrina’, and ‘SF1758’ for calibration and validation) and the other in the spring-summer 2019-2020 (only cultivar ‘SF1758’ for validation). In the autumn-winter, the relationships between individual LA (dependent variable) and L, W or L×W (independent variables) were adjusted using the linear, exponential and power models. These models were developed individually for each cultivar, as well as for the three cultivars together (universal models). In the validation between observed and estimates values, the best estimates of individual LA of cauliflower were obtained when the product L×W was used as an independent variable, being recommended the linear (LA = -14.424 + 0.843L×W) or potential [LA = 0.551(L×W)^1.057] models developed for ‘Piracicaba de Verão’ and only a linear model (LA = -22.610 + 0.928L×W) for ‘SF1758’. The universal models of the linear (LA = -13.770 + 0.833L×W) or potential types [LA = 0.578(L×W)^1.050] are recommended for estimation of cauliflower individual LA, regardless of the cultivars.