Valorisation of Three Underutilised Native Australian Plants: Phenolic and Organic Acid Profiles and In Vitro Antimicrobial Activity

Tasmannia lanceolata, Diploglottis bracteata and Syzygium aqueum are understudied native Australian plants. This study aimed to characterise the non-anthocyanin phenolic and organic acid profiles of the aqueous extracts obtained from the leaves of T. lanceolata and fruits of D. bracteata and S. aqueum by UHPLC-Q-Orbitrap-MS/MS and UHPLC-TQ-MS/MS. A total of 39, 22, and 27 non-anthocyanin polyphenols were tentatively identified in T. lanceolata, D. bracteata, and S. aqueum extracts, respectively. Furthermore, sugars and ascorbic acid contents as well as in vitro antioxidant and antimicrobial activities of the extracts were determined. Response surface methodology was applied to achieve an extract blend with a strong inhibitory effect against Pseudomonas viridiflava, the main cause of soft rot in vegetables, Bacillus subtilis, Rhodotorula diobovata and Alternaria alternata. The identified compounds including organic acids (e.g., quinic, citric and malic acids) and polyphenols (e.g., catechin, procyanidins, and ellagitannins) might contribute to the observed antimicrobial activity. Furthermore, this study provides the most comprehensive phenolic profiles of these three underutilised native Australian plants to date.


Introduction
Exploring native plants as potential sources of bioactive compounds for a range of applications in the food industry, such as preservatives (either antioxidants or antimicrobials), flavouring agents and functional ingredients, is currently of increasing interest. Australia is the native habitat for a diverse range of plant species-over 25,000-that most of them have evolved to suit the often-harsh growing conditions and have been long used by Indigenous communities for culinary or medicinal purposes. Several studies have reported the diverse phytochemical composition and health-enhancing effects of native Australian plants [1,2]. Many of them are still largely unknown, and most of them have not yet been studied for their chemical and nutritional composition as well as biological activities.
Analysis of native Australian plants may thus offer promising prospects for finding phytochemicals with strong bioactive properties. Therefore, three underutilised plants of the native Australian flora were investigated in the present study. Syzygium aqueum (Burm. F.) Alston (Myrtaceae), commonly known as the watery rose-apple or lillypilly, is native to a region ranging from tropical Asia to north Queensland (Australia), and -UHPLC, ultra-high-performance liquid chromatography; HRAM, high-resolution accurate mass. * MRM scan mode with optimized collision energy was employed for the targeted analysis and quantification of sugars and organic acids. Two multiple reaction monitoring transitions were used to quantify each sugar/acid and to confirm their identities according to their specific mass fragmentation pattern ([M-H] − → quantifier/qualifier (m/z)).

Sugar Analysis
The extraction and analysis of soluble sugars were carried out according to the method described by Hong and colleagues [13]. External calibration curves of sugar standards (2.1-260 µg.mL −1 ) were used for quantification.

Vitamin C Analysis
The extraction and analysis of vitamin C (L-ascorbic acid (L-AA) and dehydroascorbic acid (DHAA)) were carried out as described by Phan and colleagues [14]. An external L-AA (1.5-76.3 mg.mL −1 ) calibration curve was used for quantification.

Organic Acid Analysis
Analysis of organic acids was conducted as reported by Moldoveanu and colleagues [15], with some modifications. Approximately 0.5 g of extract powder was mixed with 0.2 M HCl and vortexed for 1 min, followed by sonication (15 min, 25 • C). The mixture was then shaken by a reciprocating shaker for 1 h, followed by centrifugation (3900 rpm, 10 min). The supernatant was collected, and the pellet was re-extracted two more times as described above. The supernatants were then combined and filtered (0.22 µm, PTFE). External calibration curves using a mix of organic acid standards (0.2-1030 µg.mL −1 ) prepared in aqueous formic acid (1%; v/v) were used for quantification.

Total Phenolic Content (TPC)
The TPC of the extracts was determined using the Folin-Ciocalteu method as described by Singleton et al., 1999 [16]. The results were reported as mg gallic acid equivalents per g extract.

DPPH Radical Scavenging Capacity
The DPPH radical scavenging capacity assay was carried out according to the method described by Brand-Williams and co-workers [17] with slight modifications (equal aliquots of 0.1 mM DPPH and sample). The results were reported as IC 50 (µg.mL −1 ).

HRAM Analysis and Tentative Identification of Non-Anthocyanin Phenolic Compounds
Approximately 0.5 g of extract powder was resuspended in 80% methanol containing 1% formic acid and passed through a 0.22-µm PTFE filter after centrifugation. Thirteen phenolic standards (Section 2.1) were prepared in methanol and injected into the UHPLC-MS/MS system either individually or in combination. A full MS scan in negative mode with the range of 100-1200 m/z followed by an all-ion fragmentation scan in the range of 80-1000 m/z was performed to acquire the MS and MS 2 data. The MS characteristics of each peak detected in the UV spectra were determined based on the retention time, isotope distribution of neutral mass and the MS 2 fragments spectra. Compound identification was carried out by manual comparison with injected standards (targeted identification) and matching with MS data reported in the literature and online database [18] to tentatively identify the unknown compounds (untargeted identification).
2.9. Antimicrobial Activity 2.9.1. Design of Experiments Response surface methodology using Design Expert v.11.1.2.0 (Stat-Ease Inc., Minneapolis, MN, USA) was employed to study the effect of varying extract concentrations in the blend on the inhibitory activity against the selected spoilage microorganisms and subsequently to determine the optimal extract concentrations. A 17-run Box-Behnken design consisting of five replicate centre points was developed with T. lanceolata (A, 0-10% (% is equivalent to g per 100 mL)), D. bracteata (B, 0-10%) and S. aqueum (C, 0-10%) as independent variables (Table 2). This resulted in various extract content combinations, with 25% as the highest extract content in the blend. Moreover, a 14-run randomised Simplex-Lattice mixture design with one central point was developed, and the total concentration of the extracts "%T. lanceolata (component A) + %D. bracteata (component B) + %S. aqueum (component C)" was constrained to 10% with each extract ranging from 0 to 10% ( Table 2). The effect of independent variables on the studied responses was determined through the model equations and visually expressed in 3D contour plots. A polynomial equation was used to fit the experimental data and establish the relations between the independent variables and the obtained responses. The lack-of-fit test, coefficient of determination (R 2 ), and adjusted R 2 were used to assess the validity and adequacy of the fitted model. The blend of extracts was optimised by the desirability function to maximise the inhibitory activity against the studied microorganisms.

Agar Well Diffusion Assay
Pseudomonas viridiflava, Bacillus subtilis, Rhodotorula diobovata and Alternaria alternata were taken from a culture collection of the University of Queensland (Coopers Plains, QLD, Australia), which were isolated and identified from fresh-cut capsicums (unpublished data) and stored at −80 • C. Briefly, the inoculums (10 6 CFU.mL −1 ) of overnight-grown bacteria and yeast, and 5-day-old mould were spread on Mueller Hinton (bacteria) and potato dextrose (fungi) agar plates. Three 8 mm wells were punched in the plate and filled with 100 µL of the sample. Plates were incubated at 25 • C for 48 h (fungi) and 30 • C for 24 h (bacteria). The inhibition zone diameter (mm) was measured by a digital calliper (±0.01 mm, Craftright, China) and subtracted from the well diameter. The sensitivity according to "diameter of inhibition zone" can be categorised as follows: <8 mm not sensitive, 9-14 mm sensitive, 15-19 mm very sensitive, and >20 mm extremely sensitive [19].

Statistical Analysis
All measurements were performed in triplicate, and the results were expressed as the mean value ± standard deviation. A one-way ANOVA was used to analyse the results using SPSS software (version 27; IBM Institute Inc., Armonk, NY, USA). Tukey's HSD test with a 95% confidence interval was used to compare the differences between means. Table 3 presents data on soluble sugars, vitamin C and organic acids of aqueous extracts derived from T. lanceolata leaves, D. bracteata fruits and S. aqueum fruits. As expected, more sugar and vitamin C was found in the fruit extracts than in leaves. The D. bracteata extract showed the highest total content of sugars (ca. 34 g. 100 g −1 dw) and vitamin C (2.43 mg. 100 g −1 dw), followed by S. aqueum and T. lanceolata. The variation in the sugar content is associated with differences in plant species (i.e., genetic), sun exposure due to the canopy, respiration, and ripening rates of fruits, as well as the presence and activity of specific enzymes that are involved in sugar metabolism [20]. Fructose was found to be the most abundant sugar in both fruit extracts, while sucrose was the major sugar found in T. lanceolata leaves extract. The low content of sucrose in fruit extracts can be attributed to the ripening phenomenon that causes sucrose conversion to fructose and glucose [21]. Data are mean ± standard deviation (n = 3); data with different letters in the same row are significantly different (p < 0.05). GAE, gallic acid equivalents; dw, dry weight.

Chemical Composition
To the best of our knowledge, the organic acid profile was reported for the first time for the studied extracts. Roughly seven organic acids were identified in the extracts that exhibited very different profiles (Table 3). Quinic and citric acids were the most abundant (39.38 and 38.59% of total acids, respectively) in S. aqueum extract, while malic and shikimic acids were the most abundant (84.28% and 48.05% of total acids, respectively) in D. bracteata and T. lanceolata extracts, respectively. A considerably higher total content of organic acids was detected in S. aqueum extract (74.64 g. 100 g −1 dw) compared to 26.33 g. 100 g −1 dw in D. bracteata and 11.03 g. 100 g −1 dw in T. lanceolata extracts. The concentration of organic acids in fruits and leaves depends on sugar concentrations and their use for respiration. Several studies have shown the beneficial effects of organic acids not only as antibacterial agents but also on human health, including their involvement in iron absorption, reduction of levels of circulating glucose and cholesterol, and anxiolytic effects [22,23]. Quinic acid, for example, has exhibited anti-neuroinflammatory and radioprotection effects [24], as well as anti-HIV-1 activity [25].
Our results demonstrated that the antioxidant capacity of the extracts was directly related to the total phenolic content. The TPC value of 123.47 mg GAE.g −1 dw was found in T. lanceolata leaves extract, which showed a strong antioxidant capacity (DPPH IC 50 value of 36.59 µg.mL −1 ). These results were in good agreement with those reported by Alderees and colleagues, who found 157.4 mg GAE.g −1 dw in an aqueous extract of Tasmanian pepper leaves [26]. On the other hand, the fruit extracts with low contents of TPC had a considerably low antioxidant capacity (Table 3). Unlike the fruits' extracts, leaves extracts with strong antioxidant activity can be used to reduce oxidative stress and contribute to preventing damage by reactive species. Several studies have also shown low values of TPC in S. australe (2.14 mg GAE.g −1 dw) and S. luehmannii (2.23 mg GAE.g −1 dw) [27,28]. The accumulation of phenolic compounds in different plant tissues is influenced by environmental conditions such as temperature, sun exposure and other weather conditions, which may explain the observed differences in the studied extracts. For instance, the observed higher TPC in Tasmanian pepper leaves could be attributed to the increased expression of genes associated with flavonoid biosynthesis due to high sun exposure [29]. Generally, the biosynthesis of phenolic compounds in plants is the result of a collection of regulatory signals, including developmental (e.g., during anthocyanin production during fruit and flower development) and environmental (e.g., protection against abiotic and biotic stresses) signals [30].

Identification of Non-Anthocyanin Polyphenols
Tables 4-6 present the data on untargeted screening and tentative identification of non-anthocyanin polyphenols in the aqueous extracts of T. lanceolata leaves, D. bracteata fruits and S. aqueum fruits, using HRAM-UHPLC-Q/Orbitrap-MS/MS. The retention time and MS/MS fragmentation pattern were compared with the reported data in previous studies. A total of 39, 22, and 27 non-anthocyanin polyphenols were tentatively identified in T. lanceolata, D. bracteata, and S. aqueum aqueous extracts, respectively. The UHPLC-UV chromatograms, the mass spectra data of not-yet identified compounds (due to the unavailability of commercial standards and limited MS data in the literature), commercial standards used in this study, as well as representative full-scan and product ion mass spectra, are summarised in the Supplementary Materials (Figures S1-S7, Tables S1-S4).
Hydroxycinnamic acid compounds were also tentatively identified in the D. bracteata extract. Compound 14 was tentatively assigned as a coumaric acid derivative, producing the main fragment at m/z 119.0496 [p-coumaric acid-H-CO 2 ] − [34]. Compound 17 was tentatively suggested as a caffeoyl glucose derivative that showed a product ion at m/z 341.607, most likely [caffeoyl glucose-H] − , and another at m/z 161.0609, which was reported as a caffeoyl glucose fragment [35]. Furthermore, compounds 16, 20, and 22 were tentatively assigned as cinnamic acid derivatives according to the main fragments produced at m/z 147. 0445  Three tentatively identified hydroxybenzoic acid compounds were also found in the T. lanceolata extract. Compounds 3 and 4 were tentatively assigned as protocatechuic acid-O-hexoside [36] and protocatechuic acid [37], respectively. Compound 10 may be a hydroxybenzoic acid derivative with the precursor ion at m/z 447.1867 [M-H] − that dissociated to m/z 137.0238, which corresponds to a [hydroxybenzoic acid-H] − adduct. One dihydroxybenzoic acid was tentatively identified in the D. bracteata extract as hypogallic acid (compound 4), producing m/z 108.0213 [M-H-COOH] − as the main fragment [38]. Interestingly, the only phenolic acids found in the S. aqueum extract were compounds 3, 5 and 8, belonging to the benzoic acid group, and were identified as gallic acid (confirmed by commercial standard), bergenin [39] and theacitrin A (ester derivative) [40].

Flavonoids
Flavonoids were the most abundant compounds found in the three studied extracts (Tables 4-6 Apigenin (compound 37) was tentatively identified as another flavone present in the T. lanceolata extract [42]. Compound 32 was tentatively assigned as an apigenin derivative due to the produced characteristic fragment at m/z 269.0450 [apigenin-H] − . Compound 39 fragmented into m/z 268.0371 [M-CH 3 -H] − as well as the characteristic apigenin fragments at m/z 117.0337 and m/z 151.0030, and was therefore tentatively identified as apigenin-7,4 -dimethyl ether [44].
The flavonol, quercetin, was identified in the T. lanceolata (compound 34) and S. aqueum (compound 27) extracts, which was confirmed by a commercial standard. Two more flavonols were identified in the S. aqueum extract. Compound 26 was confirmed as myricetin using a commercial standard, and compound 7 was tentatively identified as a kaempferol derivative, with the main fragment ion at m/z 284.0318 [kaempferol-H] − . Catechin (compound 8) and epicatechin (compound 12), two flavanols, were identified in the D. bracteata extract and confirmed by commercial standards. Furthermore, compound 11 in the S. aqueum extract was tentatively assigned as gallocatechin gallate, producing the two characteristic fragment ions at m/z 125.0238 and m/z 169.0137. Compound 11 in the D. bracteata extract was tentatively identified as an eriodictyol derivative, a flavanone, with fragment ions at m/z 287.0552 as a deprotonated eriodictyol adduct and m/z 125.0239 as an eriodictyol fragment [45].

Flavonoid Glycosides
Flavonoid glycosides were also detected in the T. lanceolata extract. Compound 17 was tentatively identified as catechin rhamnoside, showing the main fragment ion at m/z 289.0709 by a 146 Da rhamnoside residue loss. Compounds 21 and 22 were identified as rutin (syn: quercetin-3-rutinoside) and quercetin-3-O-glucoside, respectively, and confirmed by commercial standards. The characteristic fragment ions at m/z 300.0275 and m/z 300.0266 were produced by a 308 Da rutinose and 162 Da glucose loss, respectively. Compounds 24 and 23 were tentatively identified as vitexin/isovitexin and vitexin/isovitexin dimer, respectively. The  [32,46], and produced by a neutral loss of 308 Da (probably rhamnoglucose) and 278 Da (probably rhamnoxylose), respectively. Furthermore, compound 30 was tentatively identified as luteolin glycoside, showing luteolin aglycone as the main fragment ion at m/z 285.0402 [M-162-H] − through the neutral loss of a hexose residue such as glucose or galactose [47]; however, glucose is more likely since it is the most common hexose in nature. Compound 31 was tentatively assigned as apigenin dihexoside [48], which produced apigenin as the main fragment (m/z 269.0451) through the neutral loss of two hexosyl groups.
Two flavonoid glycosides were also detected in the D. bracteata extract. Compound 18 was tentatively identified as isorhamnetin glycoside (a flavonol glycoside) by producing the two characteristic fragment ions at m/z 299.0194 and m/z 314.0470 [49], with the latter one resulting from a hexose loss. Compound 9 was tentatively identified as a catechin glycoside by producing m/z 289.0712 (catechin) as the main fragment ion through a 158-Da loss.

Polyflavonoids
Procyanidins, also known as condensed tannins, are classified as polyflavonoids that were found in both T. lanceolata and D. bracteata extracts. Compounds 9 and 19 in the T. lanceolata extract as well as compounds 6, 7 and 10 in the D. bracteata extract were tentatively identified as B-type procyanidin dimers (or (epi)catechin-(epi)catechin) with a precursor ion at m/z 577.  [32,47]. However, the difference in their fragmentation patterns can be attributed to the differences in monomeric flavan-3-ol unit linkages, leading to different isomers' formation [57]. Compounds 18 and 20 in the T. lanceolata extract were tentatively assigned as procyanidin dimer monoglycoside, producing the diagnostic fragment ions at m/z 289.0714 (probably formed through quinone methide cleavage [58]), m/z 245.0814, m/z 587.1086 (152 Da loss through RDA fission) and m/z 569.0995 (152-Da loss with a further 18 Da loss through dehydration) [56]. Compound 2 in the T. lanceolata extract [59] and compound 13 in the D. bracteata extract [50] were tentatively identified as procyanidin trimers, showing the characteristic procyanidin fragmentation pathway. Furthermore, compound 5 (m/z 593.1262) in the D. bracteata extract was tentatively assigned as prodelphinidin A-type [50].

Tannins
Tannins, including one complex and seven hydrolysable tannins, were only detected in the S. aqueum extract.  [39]. Compound 19 was identified as ellagic acid (m/z 300.9978) and confirmed by a commercial standard. This was detected as one of the characteristic fragment ions of five tentatively identified ellagitannins found in the S. aqueum extract. Compounds 4, 10 and 13 were tentatively assigned as castalagin [60], chebulagic acid [61] and casuarinin [62]. Compound 16 was tentatively identified as an ellagic acid derivative owing to the diagnostic fragment ions at m/z 299.9907 and m/z 300.9964. The presence of fragment ions at m/z 169 and m/z 301 reveals that the hydrolysable tannin molecule contains a simple galloyl ester and a hexahydroxy-diphenoyl (HHDP) moiety [63]. The observed fragmentation pattern of hydrolysable tannins was in agreement with those found in the literature [64], showing the characteristic loss of galloyl, HHDP, HHDP-glucose, and galloyl-HHDP-glucose. Compound 9 was tentatively identified as a galloylated tannin compound showing fragment ions at m/z 125.0237, m/z 169.0139 and m/z 633.0800.

Other Polyphenols
Other tentatively identified polyphenols detected in the T. lanceolata extract were compound 5 as a hydroxybenzaldehyde derivative [35], two stilbenes, including compound 7 as a piceatannol derivative [65] and compound 11 as pelargonidin-3-pentoside [ [36]. A coumarin derivative (compound 3) with the main fragment at m/z 147.0441 corresponding to coumaric acid was tentatively identified in the D. bracteata extract. Furthermore, compound 20 was tentatively assigned as a carnosic acid derivative, as the fragment ions at m/z 331.1887 and m/z 269.0455 correspond to carnosic acid and its fragment adducts. Moreover, organic acids, including malic acid [47] and citric acid [67], were tentatively identified in the D. bracteata (compound 1) and S. aqueum (compounds 1 and 2) extracts.

Antimicrobial Activity of Extract Blends
The aqueous extracts of plant tissues are rich in various phytochemicals that are readily soluble in water and influence their antimicrobial activity. However, the inhibitory activity can be improved by mixing two or more plant extracts through the synergistic interactions between their major and minor constituents. Hence, RSM optimization through Box-Behnken and Simplex-Lattice designs was performed for the first time on T. lanceolata, D. bracteata, and S. aqueum aqueous extracts as potential natural preservatives to find two optimised blends exhibiting the highest inhibitory activity against the growth of common spoilage microorganisms in vegetables. Table 2 shows the experimental matrix designs and results using Box-Behnken and Simplex-Lattice. Quadratic polynomial equations for predicting the inhibitory activity of extract blends against P. viridiflava, B. subtilis, R. diobovata and A. alternata were determined by multiple regression analysis of the experimental data obtained from Box-Behnken (Equations (1)-(4)) and Simplex-Lattice (Equations (5)-(8)) designs. In order to fit the data to the respective models, Box-Cox transformation and stepwise model reduction were performed, if needed. The resulting equations, including the hierarchy required for insignificant (p > 0.05) and significant (p < 0.05) terms, are given below: R 7 + 0.5 = 4.54A + 0.72B + 0.71C + 7.20AB + 8.25AC R 8 + 0.5 = 2.38A + 0.67B + 1.05C + 3.28AB + 2.11AC (8) Table 7 summarises the statistical parameters obtained by performing ANOVA to check the reliability and adequacy of the developed models (details are given in Supplementary Materials, Tables S5-S12). The R 2 values were in the range of 0.89-0.99 for the Box-Behnken design and 0.77-0.99 for the Simplex-Lattice design, showing sufficient model accuracy. This indicates that not only can the fitted models explain (p < 0.05) most of the variability in the experimental data, but there is also a strong correlation between the experimental and predicted values. In addition, the low reliability of the developed models for A. alternata in the Box-Behnken design and for B. subtilis and A. alternata in the Simplex-Lattice design was indicated by low R 2 values, although they can be used to generate adequate desirability models. The insignificant p-values (p > 0.05) of the lack-of-fit test indicated that the models fit the inhibitory activity of extract blends and confirmed the reliability of the predicted models. However, the significant p-values (p < 0.05) of the lack-of-fit test for the inhibitory activity against R. diobovata demonstrated that the obtained quadratic models did not fit well in these experimental designs, even after Box-Cox transformation and stepwise model reduction, and therefore, these models cannot be used for predictions. The growth inhibitory zone of Gram-negative P. viridiflava, the major cause of soft rot in vegetables such as capsicum, was in the range of 0-20.33 mm (Box-Behnken) and 0-19.52 mm (Simplex-Lattice) ( Table 2). The most potent antibacterial extract was S. aqueum, followed by D. bracteate. In the presence of 5% S. aqueum and D. bracteata extracts, inhibition zones of 13.20 and 9.02 mm were observed, respectively, while T. lanceolata extract did not exhibit any inhibitory effect against P. viridiflava (Table 2). Increasing the concentration of extracts to 10% gave rise to an increase in the inhibitory activity against P. viridiflava by 2 and 1.5 times (S. aqueum and D. bracteata, respectively), but did not improve T. lanceolata activity. Blending the S. aqueum extract with the other two did not improve its ability to inhibit P. viridiflava growth, although it assisted in improving the antifungal activity. However, lower inhibition of P. viridiflava was observed using blends containing <5% S. aqueum (Simplex-Lattice). The linear terms in both designs were shown to affect (p < 0.05) the bacterial inhibitory activity, except for T. lanceolata extract, which had an insignificant (p > 0.05) inhibitory influence on P. viridiflava growth (Box-Behnken design). The interactions and quadratic terms of aqueous extracts indicated an inverse relationship with the bacterial inhibitory activity of the blend according to the Box-Behnken design (Equations (2) and (3)). However, the interaction terms in the Simplex-Lattice design led to a significant (p < 0.05) increase in antibacterial activity (Equations (6) and (7)). This was further illustrated in two-dimensional contour plots developed from the fitted model equations (Figure 1a-f). Both designs demonstrated the greatest impact of T. lanceolata content on the fungal inhibitory activity, with yeast being more sensitive than mould (Figures 1g-l and 2c,d). No antifungal activity was observed by S. aqueum and D. bracteata extracts alone, whereas 5% T. lanceolata extract showed an inhibitory zone of 9.17 mm against R. diobovata which was doubled by increasing the concentration to 10%, and a 5.34 mm inhibitory zone was observed against A. alternata. The highest inhibitory zones against A. alternata were 8.29 and 5.96 mm using experimental runs 3 (Simplex-Lattice) and 4 (Box-Behnken), respectively, with different extracts' combinations. This indicates the potential of the Simplex-Lattice mixture design, unlike Box-Behnken, to unfold the synergistic effect of the extracts on the blend's antifungal activity at the ratio of 2/3 T. lanceolata, 1/6 D. bracteata, and 1/6 S. aqueum. This was also confirmed by ANOVA and the interaction terms of the developed models.   Figure 1. Contour plots for the inhibition zone of extract blends against Pseudomonas viridiflava (ac), Bacillus subtilis (d-f), Rhodotorula diobovata (g-i) and Alternaria alternata (j-l) as a function of independent factors (Box-Behnken design). The observed antimicrobial activity is mainly attributed to the synergistic effect of organic acids and phenolic compounds, which has been well demonstrated [68] and is considered an added benefit of using fruit extracts as preservatives. Partially hydrophobic biphenols can bind with the microbial outer membrane and cause structural changes leading to enhanced membrane permeability, leakage of vital intracellular constituents, and disruption of metabolism [69]. The antimicrobial properties of different phenolic compounds have been extensively studied, such as catechin [70], gallic acid [71], chlorogenic acid [72], and hydrolysable tannins [73], which were identified as major phenolic compounds in the studied extracts. Procyanidin, a tentative major phenolic compound in D. ) as a function of independent factors (Simplex-Lattice design).
The observed antimicrobial activity is mainly attributed to the synergistic effect of organic acids and phenolic compounds, which has been well demonstrated [68] and is considered an added benefit of using fruit extracts as preservatives. Partially hydrophobic biphenols can bind with the microbial outer membrane and cause structural changes leading to enhanced membrane permeability, leakage of vital intracellular constituents, and disruption of metabolism [69]. The antimicrobial properties of different phenolic compounds have been extensively studied, such as catechin [70], gallic acid [71], chlorogenic acid [72], and hydrolysable tannins [73], which were identified as major phenolic compounds in the studied extracts. Procyanidin, a tentative major phenolic compound in D. bracteata and T. lanceolata extracts, has been shown to affect the strength of the lipopolysaccharide outer barrier in Gram-negative bacteria as observed by cranberry polyphenols ) as a function of independent factors (Simplex-Lattice design).
The observed antimicrobial activity is mainly attributed to the synergistic effect of organic acids and phenolic compounds, which has been well demonstrated [68] and is considered an added benefit of using fruit extracts as preservatives. Partially hydrophobic biphenols can bind with the microbial outer membrane and cause structural changes leading to enhanced membrane permeability, leakage of vital intracellular constituents, and disruption of metabolism [69]. The antimicrobial properties of different phenolic compounds have been extensively studied, such as catechin [70], gallic acid [71], chlorogenic acid [72], and hydrolysable tannins [73], which were identified as major phenolic compounds in the studied extracts. Procyanidin, a tentative major phenolic compound in D. bracteata and T. lanceolata extracts, has been shown to affect the strength of the lipopolysaccharide outer barrier in Gram-negative bacteria as observed by cranberry polyphenols ) as a function of independent factors (Simplex-Lattice design).
The observed antimicrobial activity is mainly attributed to the synergistic effect of organic acids and phenolic compounds, which has been well demonstrated [68] and is considered an added benefit of using fruit extracts as preservatives. Partially hydrophobic biphenols can bind with the microbial outer membrane and cause structural changes leading to enhanced membrane permeability, leakage of vital intracellular constituents, and disruption of metabolism [69]. The antimicrobial properties of different phenolic compounds have been extensively studied, such as catechin [70], gallic acid [71], chlorogenic acid [72], and hydrolysable tannins [73], which were identified as major phenolic compounds in the studied extracts. Procyanidin, a tentative major phenolic compound in D. bracteata and T. lanceolata extracts, has been shown to affect the strength of the lipopolysaccharide outer barrier in Gram-negative bacteria as observed by cranberry polyphenols ) as a function of independent factors (Simplex-Lattice design).
The observed antimicrobial activity is mainly attributed to the synergistic effect of organic acids and phenolic compounds, which has been well demonstrated [68] and is considered an added benefit of using fruit extracts as preservatives. Partially hydrophobic biphenols can bind with the microbial outer membrane and cause structural changes leading to enhanced membrane permeability, leakage of vital intracellular constituents, and disruption of metabolism [69]. The antimicrobial properties of different phenolic compounds have been extensively studied, such as catechin [70], gallic acid [71], chlorogenic acid [72], and hydrolysable tannins [73], which were identified as major phenolic compounds in the studied extracts. Procyanidin, a tentative major phenolic compound in D. bracteata and T. lanceolata extracts, has been shown to affect the strength of the lipopolysaccharide outer barrier in Gram-negative bacteria as observed by cranberry polyphenols ) as a function of independent factors (Simplex-Lattice design).
The observed antimicrobial activity is mainly attributed to the synergistic effect of organic acids and phenolic compounds, which has been well demonstrated [68] and is considered an added benefit of using fruit extracts as preservatives. Partially hydrophobic biphenols can bind with the microbial outer membrane and cause structural changes leading to enhanced membrane permeability, leakage of vital intracellular constituents, and disruption of metabolism [69]. The antimicrobial properties of different phenolic compounds have been extensively studied, such as catechin [70], gallic acid [71], chloro-genic acid [72], and hydrolysable tannins [73], which were identified as major phenolic compounds in the studied extracts. Procyanidin, a tentative major phenolic compound in D. bracteata and T. lanceolata extracts, has been shown to affect the strength of the lipopolysaccharide outer barrier in Gram-negative bacteria as observed by cranberry polyphenols [74]. Moreover, phenolic compounds with antioxidant activity can bind the essential growth nutrient "iron" and therefore inhibit microbial growth. Guo and colleagues observed a considerable iron binding capacity by quercetin in low pH environments [75], which is one of the major phenolics in the studied extracts, including S. aqueum, which also contains high amounts of organic acids such as quinic acid. The presence of organic acids, on the other hand, enhances the bacteria's susceptibility to phenolic sublethal damage by reducing pH in the extra/intracellular environment, causing chemical gradient collapse, and interrupting metabolic pathways [76]. In addition, several studies have reported the wide-spectrum antibacterial activity of quinic acid, citric acid, and malic acid, and their combinations, which were identified as major organic acids in S. aqueum and D. bracteata extracts ( Table 3). The presence of sugars in the extracts can also contribute to the observed antimicrobial activity, as was suggested by Lacombe and co-authors. The authors reported the effect of sugar fractions of cranberry juice on its antimicrobial activity [74], which can be attributed to the osmotic effect of sugar compounds on microbial cells. However, this needs to be further investigated. The lower antimicrobial activity against A. alternata can be attributed to the mould's phenylacrylic acid decarboxylase system for degrading high concentrations of organic acids for their spores to survive and outgrow [76]. However, the presence of chlorogenic acid, a major compound in T. lanceolata extract, can contribute to its antifungal activity. Several studies have shown the fungicidal activity of chlorogenic acid and its derivatives, which occurs through fungal cell lysis and permeabilization of the spore membrane [77]. Nevertheless, further studies are needed to understand the antimicrobial mechanism of these extracts and the role of sugars, organic acids, and phenolic compounds, as well as major and minor compounds.
The RSM desirability function was used to optimise the blends and maximise the antimicrobial activity. The two extract blends containing "9.35% T. lanceolata, 5.00% D. bracteata and 5.00% S. aqueum" and "4.72% T. lanceolata and 5.28% S. aqueum", respectively, presented the best combinations based on the models developed through Box-Behnken and Simplex-Lattice designs. The d-values for the optimised combinations were 0.927 and 0.822 (Box-Behnken and Simplex-Lattice, respectively), indicating that about 93% and 82% of desirability in statistical optimisation were satisfied. The RSM models were tested by performing an external validation using the optimised extract blends ( Table 8).
The experimental values for inhibitory activity were within the ±95% prediction limits proposed by the regression models, which confirms the reliability and predictivity of the developed models.

Conclusions
To the best of our knowledge, this study provides, for the first time, information about organic acids and non-anthocyanin polyphenols in the aqueous extracts from T. lanceolata leaves, D. bracteata and S. aqueum fruits. The potential of aqueous extracts containing various phytochemicals such as organic acids and non-anthocyanin polyphenols to inhibit the growth of spoilage microorganisms, in particular P. viridiflava, which causes soft rot in a wide range of vegetables, was also demonstrated. The results obtained in this study could suggest various value-added applications for these plant materials and their extracts. Indeed, being a high source of bioactive compounds with antioxidant and antimicrobial properties such as polyphenols, these plants could be valorised as an industrial source of bioactive compounds, which will find application as effective alternatives to conventional chemical preservatives in the food, pharmaceutical, and cosmetic sectors. However, further studies are needed to confirm the identity of the tentatively identified compounds, to assess the impact of harvest time and storage conditions on the polyphenol/phytochemical composition, and to find stronger antifungal plant extracts.