The aim of the present work has been to characterize, by NMR-based metabolic profiling, extravirgin olive oils (EVOOs) from a subarea (Salento) of Apulia, leader EVOO producer among the Italian regions. According to the European Union (EU) definition, Protected Designation of Origin (PDO) products are mostly closely linked to the concept of terroir due to the place of origin, climate and local know-how. Moreover, the authenticity and traceability of several products such as olive oils with specific geographical origin require to be preserved by analytical methods. In this regard, about a hundred EVOO samples (monovarietal and blend samples, cultivars Ogliarola Salentina and Cellina di Nardò, basis of “Terra d’Otranto” PDO, campaign 2012-2013) were therefore analyzed by 1H NMR spectroscopy and multivariate statistical analysis. Both unsupervised (PCA) and supervised (OPLS-DA) statistical analyses allowed differentiation of monocultivar oils and blends characterization. Other features such as the age of the trees (young, <100 years, and secular olive trees, >100 years) could also be investigated. Cellina samples showed a higher content of aldehydic and phenolic compounds, while Ogliarola samples were characterized by NMR signals in the range of δH 6.5 - 5.6, which could be ascribed to higher carotenoids content. Higher polyphenols and polyunsaturated fatty acid content were also found in young over secular tree EVOOs.
Olea Europaea L. (family, Oleaceae), commonly known as “olive”, is among the oldest known cultivated trees in the world and in particular the most abundant in the Mediterranean basin. The health beneficial effects of olive fruit and oil, in particular of extravirgin olive oils (EVOOs), are well known and documented [
93 authentic EVOO samples were collected during the harvesting period 2012-2013 from different microareas of Lecce province (Le, Italy): 26 monocultivar Cellina di Nardò; 32 monocultivar Ogliarola Leccese; 35 blend Cellina/Ogliarola samples (
All chemical reagents for analysis were of analytical grade. CDCl3 (99.8 atom %D) and tetramethylsilane, TMS (0.03 v/v %) were purchased from Armar Chemicals (Switzwerland).
For NMR sample preparation ~140 mg of olive oil was dissolved in deuterated chloroform (CDCl3 with TMS as internal standard) adjusting the mass ratio of olive oil:CDCl3 to 13.5%:86.5%. 600 µL of the prepared mixture was transferred into a 5 mm NMR tube. NMR spectra were recorded on a Bruker Avance III spectrometer (Bruker, Karlsruhe, Germany), operating at 400.13 MHz for 1H observation and a temperature of 300.0 K, equipped with a BBO 5 mm direct detection probe incorporating a z axis gradient coil. NMR spectra were acquired using Topspin 2.1 (Bruker). Automated tuning and matching, locking and shimming using the standard Bruker routines ATMA, LOCK, and TopShim were used to optimize the NMR conditions. Experiments were run in automation mode after loading individual samples on a Bruker Automatic Sample Changer, (BACS-60), interfaced with the software IconNMR (Bruker). Two different 1H NMR experiments were performed for each sample: a standard one-dimensional 1H ZG NMR experiment and a one-dimensional 1H NOESYGPPS NMR pulse sequence with suppression of the strong lipid signals (20 frequencies), in order to enhance signals of minor components present in EVOOs (Bruker). Spectra were obtained by the following conditions: zg pulse program (for 1H ZG NMR) 64 K time domain, spectral width 20.5555 ppm (8223.685 Hz), p1 12.63 μs, pl1 −1.00 db, 16 repetitions; noesygpps1d.comp2 pulse program (for 1H NOESYGPPS NMR) 32 K time domain, spectral width 20.5555 ppm (8223.685 Hz), p1 12.63 μs, pl1 −1.00 db, 32 repetitions.
NMR data were processed using Topspin 2.1 (Bruker) and visually inspected using Amix 3.9.13 (Bruker, Bios pin). 1H NMR spectra were obtained by the Fourier Transformation (FT) of the FID (Free Induction Decay), applying an exponential multiplication with a line-broadening factor of 0.3 Hz. The resulting 1H NMR spectra were manually phased and baseline corrected using the Bruker Topspin software. Chemical shifts were reported with respect to the TMS signal set at 0 ppm. 1H NMR spectra were segmented in rectangular buckets of fixed 0.04 ppm width and integrated, using the Bruker Amix software. Bucketing of 1H ZG NMR spectra (BUCKET-1) and 1H NOESYGPPS NMR spectra (BUCKET-2) were obtained within the range 10.0 - 0.5 ppm (BUCKET-1) and 10.0 - 5.6 ppm (BUCKET-2), respectively. In both cases, the spectral region between 7.60 and 6.90 ppm was discarded because of the peak due to residual protic chloroform signal at 7.24 ppm. The remaining buckets were then normalized to total area to minimize small differences due to total olive oil concentration and/or acquisition conditions among samples. A third data set named BUCKET-3 was generated combining BUCKET-1 and BUCKET-2 in one matrix (1 line per olive oil sample).
The potential to correlate origin of authentic olive oil samples with NMR data was studied using a combination of established multivariate statistical tools, such as unsu-
pervised (PCA) and supervised (PLS-DA, OPLS-DA) statistical techniques. Multivariate statistical analysis and graphics were obtained using Simca-P version 13.0.2 (Umetrics, Sweden) and different procedures were used: Principal Component Analysis (PCA), Partial Least Squares Discriminant Analysis (PLS-DA) and Orthogonal Partial Least Squares Discriminant Analysis (OPLSDA) [
As reported in details in the experimental section, three different bucket datasets were generated from NMR spectra: BUCKET-1 was obtained within the range 10.0 - 0.5 ppm, BUCKET-2 was obtained within the range 10.0 - 5.6 ppm and BUCKET-3 was the combination of the two previous bucket tables (taking into account only the range 5.0 - 0.5 ppm originating from BUCKET-1 and the whole BUCKET-2). For every bucket table built, multivariate statistical analyses (unsupervised, PCA, and supervised methods, PLS-DA and OPLS-DA) were applied. PCA allowed to obtain a general overview of the natural data grouping. The original datasets were rearranged in a new multivariate coordinate space where the dimensions are ordered by decreasing explained variance in the data. The principal components were displayed as a set of scores (t), which highlight clustering or outliers, and a set of loadings (p), which highlight the influence of input variables on t. In all the models studied, PCA did not show significant trends or clustering with the exception of t2/t4 PCA score plot (
overlap was observed between the two cultivars. Nevertheless, by examining the loadings of the original bucket variables, Cellina samples were characterized by variables with negative loadings on t4. In particular, signals at 9.64 and 6.64 ppm were attributed to aldehydic and phenolic compounds, respectively. On the contrary, Ogliarola samples were characterized by positive loadings on t4 of signals in the range of δH 6.5 - 5.6, which could be ascribed to carotenoids. In order to improve the separation among oils based on maximizing covariance between the measured data (X) and the response variable (Y), OPLS-DA models were also studied. By this method the identity of each group of samples is specified such that maximum variance of the groups can be attained in the hyperspace. OPLS-DA applied to the same two most representative cultivars of Salento area (Cellina di Nardò and Ogliarola Leccese) gave a good model (1 predictive and 2 orthogonal) with R2 = 0.661 and Q2 = 0.448. The predictive variation, t1, corresponds to 9.01% of all variation in the data and the uncorrelated variation, to1 (orthogonal variation), corresponds to 2.22%. The score plot showed a clear separation of the two groups (
on olive oil chemical composition highlights that the polyphenols are remarkably variable according to the variety, the agronomic conditions, the state of ripeness, and the technology of conservation [
In general, concentrations of some molecules of the unsaponifiable fraction of EVOOs (such as minor components) and fatty acids resulted significantly different for the two cultivars considered. In addition, OPLS-DA was performed on Cellina di Nardò (26 samples) and Ogliarola Leccese (32 samples) using the statistical models for classification purposes of blend samples (35 Cellina/Ogliarola samples). Interestingly, both the PCA and OPLS-DA models had a good descriptive ability. The performance classification of OPLS-DA for blend EVOOs (Cellina/Ogliarola samples) is shown in the score plot tPS[
viding that they all were obtained in the same relatively small geographical area such as for Salento EVOOs.
In any case, both unsupervised and supervised methods are required for this kind of study, in particular PCA to look for trends among samples and possible outliers, while OPLS-DA to simply interpretation of data in the case of known class information [
It is well known that Apulia region is the most important area for olive oil production in Italy, accounting for almost 40% of the total country production [
This study provides an initial evaluation of how natural variability in the olive oil might affect blends originating from specific cultivars. It is worth noting that Ogliarola di Lecce (also known as Salentina) and Cellina di Nardò, which are the basis of “Terra d’Otranto” PDO EVOOs (alone or in combination at least for 60% [
This work was supported by Apulia region project grant (PIF mis. 124 Filiera Olivicola 100% Pugliese JonicoSalentina).
We thank Agricola Nuova Generazione soc. coop. Agricola and Dr Carmelo Buttazzo for providing and organizing EVOO samples.