GC Jacobassi1, TWA Balsalobre1, NE Pinto1, RG Chapola1,2, HP Hoffmann1,2 and MS Carneiro1,2
1Sugarcane Breeding Program of RIDESA/UFSCar, Araras, SP, Brazil; monalisa@ufscar.br
2Federal University of São Carlos, Araras, SP, Brazil
The development of new sugarcane cultivars requires a systematic selection process to identify genotypes with desirable traits, such as tillering, disease resistance, sucrose content, and high yield. Near-Infrared (NIR) spectroscopy has been integrated into breeding programs to enhance the accuracy of sugar content measurements. Between 2020 and 2023, a calibration curve for sucrose content was developed using approximately 50 samples collected from Brazilian Panel of Sugarcane Genotypes (BPSG) and additional samples from genetic breeding trials conducted by RIDESA/UFSCar, Brazil. Spectra were obtained from the middle third of the stalks of each genotype. Conventional laboratory analyses followed the CONSECANA protocol to determine reference values for sucrose content of the juice. Spectra data were preprocessed using the standard normal variate and the Savitzky-Golay first derivative. Spectral outliers were identified through principal component analysis and Mahalanobis distance. A partial least squares regression was used to construct the calibration curve. Model performance was evaluated using the coefficient of determination (R²) and root mean squared error (RMSE). The results demonstrated a calibration dataset R² of 0.93 and an RMSE of 0.87%. This non-destructive approach offers a faster alternative to traditional sucrose estimation methods, particularly useful in both early and advanced stages of sugarcane breeding programs.