Tracing autochthonous pig breeds with meat near-infrared spectra data pig

Near Infrared spectroscopy (NIRS) offers an easy to use and cost-effective tool for quantitative and qualitative application in animal science. Of special interest could be considered the use of NIRS for tracing meat origin on the basis of breed specification. The aim of this study was to investigate the potential use of meat NIRS as a “fingerprint” of autochthonous pig breeds. The research considered intact and grounded sample of Longissimus Dorsi (n=371) collected from 11 European local pig breeds from the TREASURE project, namely: Alentejana (ALE), Bisara (BIS), Crna Slawonska (CRN), Gascon (GAS), Iberian (IBE), Krskopolje (KRS), Lithuanian Wattle (LIA), Lithuanian White (LIH), Negre Mallorqui (NEG), Schwabish Hallisches (SCH) and Turopolje (TUR). For each muscle sample, two aliquots were scanned using FT-NIRS Antaris II model (Thermo Fisher Scientific) in absorbance mode considering the infrared region (3999 to 9999 cm-1) and averaged. Discriminant analysis of principal components (DAPC) on meat NIRS was used to assess: i) breed traceability and ii) similarity among breeds. DAPC was applied on standardised (centred and scaled) spectra using the R package. For breed traceability, cross-validation was applied: five samples per breed were sampled at random, without replacement, and used in validation. The procedure was repeated ten times and each breed was analysed separately. In this case, all breeds were present in the training set. Similarity among breeds was assessed by excluding each breed from the training set and assigning the samples in the validation set to the breeds in the training one. Overall correct classification was 68.0 and 77.6% for intact and grounded meat, respectively. Alentejana had 100% correct classification for both intact and grounded meat. For CRN, KRS, LIA, LIH and NEG use of grounded meat spectra resulted in higher classification rates from 44 to 64% for intact and from 66 to 90% for grounded meat), while for GAS the opposite was found (80% for intact vs. 70% for grounded). For the rest of the breeds slight or no differences were observed between intact and grounded samples and classification rates ranged between 66 (CRN and SCH) to 72% (BIS). The lowest classification rates were observed in both cases for CRN. Similarity among breeds was greatly varied upon dataset used (e.g. ALE samples were classified as IBE (40%), TUR (40%) and SCH (10%) using intact meat while 100% were classified as CRN using grounded meat). Our results mark NIRS as a promising tool for traceability of pig breed meat origin and support the use of grounded over intact samples.