*3.2. Statistical Analysis*

Some of the individual phenolic compounds present in the EVOOs extracts were identified and quantified by UPLC. However, the choice to evaluate the antibacterial activity of the entire extracts was taken for different reasons. First, the antibacterial activity of phenolic compounds is generally well-known [26–31]. Moreover, PF extracts might exhibit more beneficial effects than their individual constituents, which can change own properties in the presence of other compounds present in the extracts [32]. As said by Liu [33], the health benefits of fruits and vegetables give rise from synergistic effects of phytochemicals and the advantages on human health of a diet rich in fruits and vegetables is attributed to the complex mixture of phytochemicals present in whole foods. This explains why generally no individual antibacterial effect can substitute the combination of natural phytochemicals to achieve the health benefits [34]. Thus, we statistically correlated the total polyphenols and individual molecules to the antibacterial activity exhibited by the EVOO extracts. The correlation between total polyphenols and the average antibacterial activity resulted high (=0.85). We identified 10 polyphenols through UPLC analysis, based on the retention time of corresponding standards. For all of them, we calculated the percentage present in each extract. Data on polyphenol composition are reported in Table 3. The statistical approach allowed us to divide such molecules into different groups, with respect to their potential influence on the average antibacterial activity of the extracts. Correlation coefficients (Corr-coeffs) are reported in Table 4. In the first group, we found that flavonol quercetin

and isoflavone formononetin, which Corr-coeffs (0.94 and 0.97, respectively) seemed to let us foresee by the whole their highest influence on the antibacterial activity with respect to the other molecules. Other two polyphenols, flavanone naringenin and the secoiridoid oleuropein exhibited lower Corr-coeffs (0.55 and 0.47, respectively).

Taking into account the percentage of the two molecules in the extracts, it is possible to hypothesize for this other group a little bit of predominance of correlation between oleuropein and the average antibacterial activity of the 'Ravece' extract (Figure 1, left) and between naringenin on the average antibacterial activity exerted by the 'Ogliarola' extract (Figure 1, right).

**Table 3.** Polyphenol composition, obtained by Ultra Pressure Liquid Chromatography (UPLC), of the three PF extracts of *Ogliarola*, *Ravece* and *Ruvea antica* EVOOs. The data are reported as percentage of total polyphenols.


**Table 4.** Correlation coefficients between the potential average antibacterial activity and polyphenols identified in the extracts of *Ogliarola*, *Ravece* and *Ruvea antica* EVOOs. The analysis was elaborated with respect to the percentage of each molecule present in the extracts and in an independent way with respect to the pathogens.


The correlation between another group of polyphenols and the antibacterial activity of the extracts was still less strict; thus, flavone luteolin (Corr-coeff = 0.37) and the hydroxycinnamic *p*-coumaric acid (Corr-coeff = 0.33) seemed to break the antibacterial activity of the extract *Ogliarola*. Concurrently, isoflavone dadzein (Corr-coeff = 0.28) and flavonol spiraeoside (Corr-coeff = 0.27) did not seem to enhance that of the extract *Ravece*. The other flavone apigenin exhibited a negative coefficient of correlation (Corr-coeff = −0.34). This metabolite is a known antibacterial compound [34,35]. However, in some cases its effect could be nil against some microorganisms [36].

**Figure 1.** Average antibacterial activity exerted by the three PF extracts vs. oleuropein (**left**) and vs. naringenin (**right**). On X it is reported the amount (in μg) of the molecules present in 2.5 and 4.9 μg of the PF extracts tested.

The statistical approach was also applied to evaluate the correlation between the singular molecules and the antibacterial activity with respect to the microorganisms. Table 5 reports the coefficients of correlation.

**Table 5.** Correlation coefficients between the potential antibacterial activity and polyphenols identified in the extracts of 'Ogliarola', 'Ravece' and 'Ruvea antica' EVOOs, with respect to different pathogens. The analysis was elaborated with respect to the percentage of each molecule present in the extracts, taking into account the amounts (2.5 μg and 4.9 μg) of the extracts used to determine the antibacterial activity of the extracts against different pathogens. BC: *Bacillus cereus* (strains DSM 4313 and DSM 4384); EC: *Escherichia coli*; LI: EF: *Enterococcus faecalis*; *Listeria innocua*; SA: *Staphylococcus aureus;* PA: *Pseudomonas aeruginosa*.


With respect to the strains used in the agar diffusion test, we could suppose a noticeable inhibitory effect of formononentin and quercetin against *B. cereus*. In fact, both strains of *B. cereus* (DSM 4313 and DSM 4384) seemed to be strongly inhibited by the presence of these two metabolites (Corr-coeffs = 0.97 and 0.95, respectively); concurrently, quercetin seemed to prevent the bacterial growth too (Corr-coeffs = 0.96 and 0.93, respectively). A similar effect was hypothesized against *E. coli* (Corr-coeffs = 0.94 and 0.90, respectively) and against *E. faecalis* (Corr-coeffs = 0.91 and 0.75, respectively). Thus, for instance, if formononentin seemed to confirm its influence also against *Ps. aeruginosa* (Corr-coeff = 0.95) and *L. innocua* (Corr-coeff = 0.91), on the other hand the effect of quercetin versus these two microorganisms seemed to be less effective (Corr-coeffs = 0.74 and 0.77, respectively). Therefore, other studies demonstrated a limited inhibitory effect of quercetin against *Ps. aeruginosa* [37]. A potential inhibitory effect exhibited also by luteolin (Corr-coeff = 0.73) and *p*-coumaric acid (Corr-coeff = 0.66) against *Ps. aeruginosa* was observed indeed. At the same time

naringenin (Corr-coeff = 0.78), luteolin (Corr-coeff = 0.59) and *p*-coumaric acid (Corr-coeff = 0.58) would concur in influencing, although with minor efficacy, the potential antibacterial activity of the extracts against *L. innocua*. The potential behavior exhibited by metabolites on the antibacterial activity-hypothesized through such approach- seemed to be completely different when we considered *S. aureus*. In fact, by the analysis of correlation coefficients we could hypothesize that other metabolites in place of formononentin and quercetin may have contributed to the antibacterial activity of the extracts, in particular spiraeoside, dadzein, and catechin (Corr-coeffs = 0.91; 0.90 and 0.80, respectively). Moreover, this was the unique case in which oleuropein (one of the most important and known metabolites characterizing the EVOO polyphenols) seemed to have contributed to the antibacterial activity of the extracts (Corr-coeff = 0.89). Therefore, oleuropein as well as 3-hydroxytirosol (which in our case showed a correlation coefficient of 0.84) have antibacterial activity against *S. aureus*, as demonstrated by Bisignano et al. [38]. Concurrently, statistics confirmed the controversial behavior exhibited by 3-hydroxytirosol that was active against *S. aureus* but had lower effect (Corr-coeff = 0.47) against *E. coli*, corroborating the indications given by other studies [39]. The fact that the *Ravece* extract did not contain dadzein might suggest that such metabolite in particular affected the resistance of *S. aureus*. In fact, as shown in Table 2, the MIC *Ravece* extract *versus S. aureus* was higher than 10 μg and much lower in the case of the other two extracts. The absence of catechin, which gave a correlation coefficient of 0.80 and the concurrent presence of luteolin (6.22% in *Ravece*, Corr-coeff = −0.76) could have contributed to its higher MIC value. Concomitantly, the presence of apigenin found only in the *Ruvea antica* extract with the most negative coefficient of correlation (= −0.75) would seem to support its influence on the resistance of *E. faecalis* versus that extract, as indicated by the MIC value and by the results of the inhibition zone test.
