Multivariate modelling techniques applied to metabolomic, elemental and isotopic fingerprints for the verification of regional geographical origin of Austrian carrots

An exploratory study for verifying regional geographical origin of carrots from specific production regions in Austria (Genussregionen) was performed by combining chemical fingerprinting methods, namely n(Sr-86)/n (Sr-87) isotope amount ratios, multi-elemental and metabolomic pattern. Chemometric classification models were built on individual and combined datasets using (data-driven) soft independent modelling of class analogies and (orthogonal) projections to latent structures-discriminant analysis to characterise and differentiate carrots grown in five regions in Austria. A predictive ability of 97% or better (depending on the classification technique) was obtained using combined Sr isotope amount ratios and multi-elemental data. The use of data fusion strategies, in particular the mid-level option (fusion of selected variables from the different analytical platforms), allowed highly efficient (99-100%, except soft independent modelling of class analogy with 97%) and correct classification of carrot samples.

Automated summary

This exploratory study verified the regional geographical origin of carrots from specific production regions in Austria by combining chemical fingerprinting methods, achieving a predictive ability of 97% or better through the use of data fusion strategies. The chemometric classification models were able to efficiently and correctly differentiate carrot samples grown in different regions.