Developing a fast and accurate model to estimate allometrically the total shoot leaf area in grapevines

Ten different allometric models for estimating leaf area of the primary shoots, lateral shoots and total shoot leaf area of grapevines were developed using data collected from nine different Vitis vinifera L. cultivars. Firstly, the leaf area of main or lateral shoots was estimated according to two allometric models (Lopes and Pinto; Spann and Hereema) already in use for grapevines and other fruit tree species, and three other models appositely developed in this work and obtained after slight modifications of the previous ones. Secondly, five new allometric models were hereby developed to estimate directly total leaf area of the entire vine shoot (main shoot plus laterals). Rigorous statistical analysis enabled the evaluation of the models' performance in relation to (a) the type and number of predictors used, and (b) the shoot category of interest (i.e. leaf area of main, laterals and total leaf area of the vine shoot). This allowed selecting, in each analysis, the allometric model with the best performance. Although the suitability of previously developed allometric models for estimating the leaf area of single shoots (main or laterals) was confirmed, the current research showed that these models can be improved using power functions (r improved from 0.85 to 0.88 and from 0.97 to 0.98 for main and lateral shoots, respectively). However, using these models to estimate the leaf area of an entire shoot requires separate estimation of the main shoot leaf area, that of each of its laterals and then the sum all these estimates. Considering that the number of laterals on each main shoot can be high, this can be very time-consuming. In this study, we developed, for the first time, models that can predict accurately (r range 0.95-0.98) and directly the leaf area of the entire shoot (main plus its laterals) using a limited number (3-5) of predictors.