Summary
Mutton substitution for Iamb is an occasional but significant problem within the sheep industry. The aim of this study was to determine whether chemical analysis could be used to differentiate mutton from lamb. Samples of subcutaneous fat from over the chump were collected from approximately five hundred sheep carcasses at abattoirs in New South Wales, Victoria and Western Australia. The carcasses were from sheep differing in age (mutton, hogget and lamb), sex (castrates and female), breed (Merino, Dorpa Lee with various crossbreeds) and nutrition (grain, lucerne, lucerne mixed, native pasture, pasture, pasture plus supplement and saltbush). The levels of three branched chain fatty acids (BCFAs), 4-methyloctanoic (MOA), 4-ethyloctanoic (EOA) and 4-methylnonanoic acids (MNA), were determined. Statistical modelling showed that, with pre-slaughter nutrition in the model as a random term, the BCFA levels could be used for discriminating the age of sheep as the fat samples from lamb carcasses had a lower concentration of MOA and EOA, and a higher concentration of MNA, in comparison to hogget and mutton (P < 0.05). When nutrition was excluded as a random effect from the statistical model, the MOA and MNA levels did not differentiate between lamb, hogget and mutton whereas, for EOA, lamb had a lower concentration then mutton (P < 0.05) with hogget having intermediate levels between the two. Additionally, an interaction existed between age with sex (P < 0.05) where only the female lambs had a lower concentration of EOA relative to the mutton, but this was not the case for castrates. As the nutritional history of sheep is rarely available at slaughter, the levels of EOA, MNA and MOA in fat collected from a sheep carcass cannot be used as a blind test for differentiating lamb from mutton.
Three other compounds (g-dodecalactone, d-tetradecalactone and one tentatively identified as a C17:1 hydroxy acid) were also evaluated for their potential in discriminating sheep on age but, as for the BCFAs, nutrition impacted on the measured levels of these compounds. Thus, measurements made using these compounds could not be recommended as a blind test for differentiating lamb from mutton.
As an alternative approach to using BCFA levels to distinguish sheep age, five statistical classification algorithms (support vector machines, naïve Bayes, recursive partitioning, random Forests, and random Forests with linear discriminant analysis) were applied to the total measured fatty acid profile to test their suitability to discriminate lamb from mutton. Random Forests with linear discriminant analysis was very successful in classifying the two sheep types with no false positives, suggesting that the technique could be suitable for detecting lamb substitution by mutton.
