**3. Results and Discussion**

Figure 1 shows spectra obtained at 400 and 60 MHz for the same aqueous extract, respectively. Given its higher resolution, the spectrum obtained at 400 MHz allows for the identification of most protons from the species of interest. On the other hand, several of these signals appear overlapped at 60 MHz, making the initial assignment of resonances a harder task that requires technical know-how.

However, the signals corresponding to the sugar anomeric protons and citric acid methylene protons of interest are in relatively uncluttered regions of the spectrum, and therefore their identification and quantitation is achievable. Indeed, if the selection of the integration ranges is rigorous and consistent with those employed at the high field, the integration of the signals corresponding to anomeric protons of sucrose, glucose, and fructose, as well as the citric acid methylene protons, allows us to apply the methodology developed previously [25] to predict the acceptability of the mandarin samples (Table 1).

**Table 1.** Results of the sensory evaluation (acceptability), sweetening power/citric acid ratio, predicted acceptability using the model and RMSE of the prediction of each model (60 and 400 MHz data).


\* The reported values correspond to the average of 5 replicates. The RSD was less than 20% for all the varieties analyzed.

**Figure 1.** Comparison of 1H spectra of the aqueous extract of mandarin variety B475B obtained with 400 and 60 MHz spectrometers ((**a**) and (**b**), respectively). Resonances employed in the estimations are annotated in the 400 MHz spectrum. The grayed-out region in both spectra corresponds to the residual HDO peak.

The sweetness/citric acid ratio of the samples determined at the two frequencies considered had high correlation (R2 > 0.99, Figure 2), showing the equivalence of both systems and their fitness for the intended purpose of the method.

It is then possible to study the correlation between the acceptability of the mandarin samples determined by consumers and the sweetening power/citric acid ratio obtained using the 60 and 400 MHz systems (Figure 3). Using this set of data, a linear regression model with an R2 of 0.94 and an RMSE of 0.35 was obtained using data recorded at 400 MHz. The corresponding regression parameters of the linear model derived using sugar and citric acid concentrations determined with the 60 MHz instrument were 0.96 and 0.29, indicating that acceptability prediction models of similar quality were obtained regardless of the instrument employed in their development.

In addition, the correlation between the predicted acceptability using both models was very high (R2 > 0.99), further proving the equivalence of the models derived from the two instrumental systems (Figure 4).

It is worth pointing out that although models derived from data at 400 and 60 MHz are of the same predictive quality, special attention is needed when identifying and integrating data in the low-field instrument. As stated earlier and shown in Figure 1, there is considerable signal overlap in the 3.00 to 4.30 ppm region and expertise is required to assign peaks and process these spectra accurately.

**Figure 2.** Correlation between the measurement of the sweetness/citric acid ratio of the samples obtained at 400 and 60 MHz.

**Figure 3.** Correlation between consumer acceptability and sweetening power/citric acid determined using 60 and 400 MHz data (blue triangles and orange circles, respectively).

Conversely, and due to the lower resolution of low-field instruments, the variations in the chemical shifts of sugar signals with pH have less impact on spectra recorded at 60 MHz [30]. This makes spectral referencing and alignment simpler in these instruments.

**Figure 4.** Correlation of the predicted acceptability using the models obtained with both instruments.

#### **4. Conclusions**

As demonstrated above, low-field NMR systems can be employed in the development of consumer acceptability prediction models that have identical quality to those derived from high-field NMR data. The lower purchase and running costs of benchtop spectrometers makes these chemometric-based tools more accessible for routine inclusion in fruit breeding programs, such as the Uruguayan Programa Nacional de Investigación en Producción Citrícola. Furthermore, the continuing advances in benchtop NMR instruments, which include the implementation pure shift pulse sequences, solvent suppression techniques, and multidimensional and multinuclear methods, will facilitate their application to other fields of food analysis and metabolomics.

**Author Contributions:** Conceptualization, S.D.K. and H.H.; methodology, I.M.; formal analysis, I.M.; investigation, I.M.; resources, F.R. and S.D.K.; data curation, I.M.; writing—original draft preparation, I.M.; writing—review and editing, G.M., S.D.K. and H.H.; visualization, I.M.; supervision, G.M., S.D.K., H.H.; project administration, S.D.K. and H.H.; funding acquisition, S.D.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the International Atomic Energy Agency (IAEA, award CRP D52042, "Implementation of Nuclear Techniques for Authentication of Foods with High-Value Labelling Claims") and the Agencia Nacional de Investigación e Innovación (ANII, award POS\_NAC\_2020\_1\_164240, "Estudio de Perfiles Metabólicos como Herramienta para la Caracterización de Mezclas de Compuestos Bioactivos"). The APC was covered by IAEA and the Programa de Desarrollo de las Ciencias Básicas (PEDECIBA).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** The authors would like to thank the staff at INIA Salto Grande and Área de Evaluación Sensorial (Departamento de Ciencia y Tecnología de Alimentos (CYTAL), Facultad de Química, Universidad de la República).

**Conflicts of Interest:** The authors declare no conflict of interest.
