Sharma, SanjulaMundi, Harshdeep Kaur2023-12-112023-12-112022Mundi, Harshdeep Kaur (2022). Development of near-infrared reflectance spectroscopy calibration equations for phytochemical profiling of Brassica juncea (Unpublished M.Sc. thesis). Punjab Agricultural University, Ludhiana, Punjab, India.https://krishikosh.egranth.ac.in/handle/1/5810202565Near-infrared reflectance spectroscopy (NIRS) with chemometrics has emerged as an effective technique for simple, rapid, inexpensive, safe, authentic and non-destructive determination of different biochemical compounds in various food and agricultural products. It quantifies organic matter in the sample by recording the absorbance signals in the near-infrared light for the major chemical groups in less than a minute. Considering the advantages of NIRS, the present study was conducted to develop NIRS-based prediction model as a speedy alternative for measuring quality traits, otherwise assessed using chemical methods which are often timeconsuming labor-intensive and with dangerous effects on health. Brassica crop is well known for its health promoting bioactive compounds. The precise and rapid assessment of these bioactive compounds found in Brassica juncea L. (Indian mustard) may help in determining the genetic variation for these traits across diverse oilseed Brassica germplasm. This study utilizes seeds from 178 diverse B. juncea accessions harvested in the years, 2019-20 and 2020-21 which were estimated for oil content, fatty acid profile, tocopherols and phytosterols using wet chemistry analytical methods. Simultaneously, the spectra of the same samples were acquired as the logarithm of reciprocal of reflectance (log 1/R) in the entire UV-NIR wavelength range of 400 to 2498 nm of electromagnetic spectrum by scanning on NIR System model 6500, Inc., Laurel, MD, USA. Modified partial least-square (MPLS) regression method based NIRS models were developed, wherein the model exhibited higher coefficient of determination (RSQ) and coefficient determination of cross validation (1-VR) value ranging from 0.650 to 0.987 and 0.612 to 0.981 for all the traits except stearic acid, linolenic acid and phytosterol. The traits with high RSQ also possess lowest values of standard error of calibration (SEC), and standard error of cross-validation (SECV) ranging from 1.242 to 18.428 and 1.427 to 22.033. External validation of the developed models yielded high RSQ value for oil (0.693), eicosenoic acid (0.716), erucic acid (0.870), αtocopherol (0.711), γ-tocopherol (0.829) and total tocopherol (0.725) with standard error of prediction (SEP) ranging from 1.496 to 6.166. The statistical results obtained demonstrated the efficacy of newly developed NIRS model for accurately predicting the oil, fatty acids (eicosenoic acid, erucic acid) tocopherols (α-, γ-, total tocopherol) contents. The outcomes of the study will assist in precise screening and selection of superior quality lines in Brassica breeding programs.EnglishThesis