Development of multi-cultivar models for predicting the soluble solid content and firmness of European pear (Pyrus communis L.) using portable vis-NIR spectroscopy

The feasibility of a multi-cultivar model's development for the on-site determination of the internal qualities of European pear (Pyrus communis L.) cv. 'Abate', 'Cascade', 'Conference', 'Red Comice' and 'Wujiuxiang' was investigated. In total, 240 pear samples, 48 for each cultivar, were collected during fruit post-ripening period under retail stage conditions to explore the performance of a technique for developing soluble solid content and firmness models with partial least squares (PLS) and multiple linear regression (MLR) methods A commercially available portable visible and near-infrared spectrometer was used to collect the absorbance spectra (500-1010 nm) of pears. A hybrid method, orthogonal signal correction (OSC) combined with other spectral pre-process methods, such as standardization, second-order derivation and Savitzkye-Golay smoothing, was employed to filter out the variations in signals caused by different pear cultivars. Compared with the routine PLS and MLR models, the performances of the OSC-PLS and OSC-MLR models resulted in higher ratios of prediction to deviation. A new external sample set obtained at the retail stage was used to assess the performances of the best multi-cultivar models of soluble solid content and firmness, which resulted in greater coefficients of determination for prediction of 0.87 and 0.66, respectively, and lower root mean square errors of prediction of 0.45% and 8.11 N, respectively. The results highlighted the potential of portable Vis-NIR instruments for the on-site testing of internal quality indices of European pears. The development of a multi-cultivar model for fruit is necessary and significant, in particular for improving the performance of the firmness model.