Falling number of soft white wheat by near‐infrared spectroscopy: A challenge revisited. Issue 3 (19th April 2018)

• Record Type: Journal Article
• Authors: Delwiche, Stephen R.; Higginbotham, Ryan W.; Steber, Camille M.
• Is Part Of: Cereal chemistry. Volume 95:Issue 3(2018)
• Page Start: 469
• Page End: 477
• Publication Date: 2018-04-19
• Subjects: falling number -- near‐infrared spectroscopy -- wheat; Chemistry, Technical -- Periodicals; Baking -- Periodicals; Baking; Chemistry, Technical; Edible Grain; Food Analysis; Electronic journals; Periodicals; 664.72
• DOI: 10.1002/cche.10049
• Languages: English
• ISSNs: 0009-0352

Abstract

Abstract : Background and objectives: Wheat Hagberg falling number (FN) is a long‐standing quality test that, by means of measuring the viscosity of a heated water‐meal or water‐flour mixture, characterizes the activity of endogenous α‐amylase, the enzyme primarily responsible for starch hydrolysis. The accuracy, time requirement, and cost of this test have come under heightened scrutiny, particularly in seasons when weather conditions have been favorable to preharvest sprouting or late maturity amylase. Near‐infrared (NIR) spectroscopy, an analytical approach routinely used in the grain industry to measure contents of protein and moisture, was reexamined as a possible alternative to the FN procedure. Findings: Partial least squares (PLS) regression quantitative models developed on a genetically diverse set of Washington grown white wheat demonstrated low accuracy, with standard errors of performance ranging from 40 to 77 s. Alternatively, linear discriminant analysis and PLS discriminant analysis (PLSDA) qualitative models, developed and tested using a FN cutoff (pass/fail) value, also demonstrated low accuracy, with the best model correctly identifying 67% and 71% of the samples, respectively, above and below a threshold value established as the median value of FN in a calibration set of several hundred samples. Conclusions: Replacement of the FN test with one based on NIR spectroscopy on either whole grain or ground meal for making decisions on segregating wheat lots according to α‐amylase activity is not recommended. Significance and novelty: Because NIR spectroscopy is not sufficiently accurate to quantitatively model FN or differentiate low from high FN grain, viscometry procedures for starch integrity, such as FN, will continue their use in grain commerce.