Assessment of contamination of wool by dark and medullated fibres using near infrared spectroscopy

Abstract

Increased crossbreeding of Merino sheep in Australia, involving coloured or highly medullated sire breeds, has increased the risk of dark and highly medullated fibres in wool lots. Commercial implementation of the Dark and Medullated Fibre Risk Scheme, based on producer information, is identifying to buyers some of these risks and technology is sought to provide an inexpensive method for routine presale testing of greasy wool lots. One measurement concept assessed the ability of near infrared spectroscopy (NIRS) to predict variation in levels of pigmented fibres or highly medullated fibres in wool. The project used either ‘seeded’ wool samples or naturally contaminated samples with measured reference values as well as different methods of sample preparation of wool fibre (in air or immersed in benzyl alcohol) or the solutions from alkali hydrolysis of wool fibre and NIRS measurement (reflectance v. transmission). NIRS can predict variation in trace levels of pigmented fibre or highly medullated white fibres (kemp) in wool and, among the methods assessed, reflectance analysis of wool fibre in air was generally better than the other options considered. Calibration models for NIRS reflectance measurement of 5 g wool samples ‘seeded’ with 1–50 black-pigmented, tan-pigmented or white kemp fibres gave coefficients of determination (R2) of 0.96, 0.88 and 0.82 with standard errors of cross-validation (SECV) of 4.34, 6.97 and 7.17 fibres per 5 g sample, respectively. However, these calibration equations generally did not predict variations in the reference values for 3 other sets of naturally contaminated samples. New calibration equations determined for each of the sets of naturally contaminated samples also predicted variation in the pigmented fibre reference values, with the extent of agreement depending on the accuracy of the reference data as well as sample preparation and method of measurement. Calibration models for NIRS reflectance measurement of wool fibre from the 3 sets of naturally contaminated samples produced R2 = 0.99, 0.71 and 0.92 with SECV = 0.923, 6.43 and 4.54 pigmented fibres per 5 g sample, respectively. However, these calibrations and those obtained from various combinations of the wool sets also had limited ability to predict variation in pigmented fibre reference values in other independent or excluded samples. Refinement of the technique and development of calibrations with extensive and reliable reference data, representing all of the wool variation likely to be encountered, may allow this NIRS potential to become relevant in the presale testing of wool as an inexpensive measurement procedure for estimating dark and medullated fibre content.