Impact of Application of Abscisic Acid, Benzothiadiazole and Chitosan on Berry Quality Characteristics and Plant Associated Microbial Communities of Vitis vinifera L var. Mouhtaro Plants

: The phenolic proﬁle of the grape berries is a key quality factor for the red grapevine varieties and several techniques have been applied to improve it. An innovative technique is the application of resistance elicitors and phytohormones. In the present study, leaves and berries of a Greek red indigenous variety (Mouhtaro) sprayed with two elicitors, benzothiadiazole and chitosan and a plant hormone abscisic acid, during veraison. Physicochemical and phenolic characteristics of the berries and microbial communities of rhizosphere, phyllosphere and carposphere were analyzed at harvest. Differences in the microbial communities on different plant compartments were observed after the application of the plant activators. Chitosan treatment increased the abundance of the beneﬁcial lactic acid bacteria, while the abscisic acid treatment decreased the presence of spoilage fungi on the carposphere. Treatments differentiate total phenolics, anthocyanins and in the chemical characteristics of grape must with chitosan treated grapes had increased anthocyanins and skin-derived phenolics that correlated positively with the microbial taxa that was discriminant by LefSe analysis. This research provides an overview of the effect of plant activators on the microbial ecology and grape quality of the Greek variety Mouhtaro and presents the potential of new and innovative approaches in the ﬁeld of sustainable viticulture.


Introduction
Recently the European Commission announced the "European Green Deal", a set of policy initiatives that aim to reform several European economic activities to be climate friendly by 2050 [1]. Among them, the "From Farm to Fork" strategy [2] highlights the need of food production chain sustainability, by decreasing the use of agrochemicals (fertilizers and pesticides) and, at the same time, ensuring the production of healthy, high quality, costeffective food products for the consumers. Under this framework, studies on sustainable cropping systems aim to deliver innovative solutions and provide sufficient, safe, nutritious, healthy, and affordable food. Under that scope, there is a growing interest on molecules of biological or chemical origin, like elicitors, biostimulants and plant hormones with multiple modes of action, e.g., (i) acting against pests/pathogens, (ii) increasing nutrient microbial reservoir for plants, and usually, microbial diversity is higher in the rhizosphere than at aboveground parts, due to the more uniform micro-environmental conditions, the highly selective nutrient-poor conditions and high exposure to variable abiotic factors (i.e., temperature, humidity, and UV radiation intensity) of the phyllosphere and carposhere [30].
Plant-associated microbial communities are in constant communication, controlling for, and controlled by, the biosynthetic pathways of the plant [34,35]; thus, exogenous application of plant biosynthetic activators, such as elicitors, biostimulants and phytohormones, can bring modifications to these microbial communities. Thus far, little attention has been given to fungi and bacteria, which interact with the grapevine to trigger the secondary metabolism and promote the biosynthesis of phenolic compounds. Cappelletti et al. [36] observed that fungal and bacterial communities of the phyllosphere of the grapevine cultivar Pinot noir ENTAV115 were altered after the application of nutrient broth as a plant resistance inducer. Villegas et al. [37] recorded that pectin-derived oligosaccharide acquired from the Aspergillus niger increased the color and anthocyanin content of grape berries. Moreover, arbuscular mycorrhizal fungi and plant growth-promoting pseudomonads can increase anthocyanin concentrations in strawberry fruits [38]. In addition to this, the endophytic fungi Altenaria can alter anthocyanins in grape cells [39].
In the present study, we investigated the effects of foliar application of two elicitors, chitosan and benzothiadiazole, and a plant hormone, abscisic acid, on a local red grape variety Mouhtaro cultivated from the Muses Valley (Ascri, Viotia, Greece), by analyzing berry quality characteristics, as well as plant-associated microbial communities. To address this, we monitored berry quality characteristic at harvest in parallel with the bacterial and fungal communities of the rhizosphere, phyllosphere, and carposhpere.

Experimental Set up and Sampling
Experiments were carried out in a Vitis vinifera L. cv Mouhtaro non-irrigated vineyard of Muses Estate Winery at Muses Valley (PGI Thiva, Greece; 38 • 19 30.9 (N) and 23 • 05 37.7 (E)). Vines were planted in 2007 onto a R110 rootstocks (planting density 2.5 × 1.2 m) and trained on double cordon. The pruning system followed was 3 spurs in each cordon. Vineyard management was uniform within the whole experimental plot and in compliance with the recommended agricultural practices for the given viticulture location. Three different treatments were applied. Vines were sprayed with an aqueous solutions of (i) 400 mg/L abscisic acid (s-abscisic acid 10.4% w/v, Protone SL, Hellafarm, Greece) (Treatment ABA), (ii) 0.3 mM benzothiadiazole (benzo-(1,2,3)-thiadiazole-7-carbothioic acid S-methyl ester, BTH, trade name Bion, Syngenta, Basilea, CH) and (iii) 0.3% chitosan (chitosan hydrochloride 3% w/w, CHT, Project One, Phytorgan S.A., Greece). In BTH and CHT treatments, Tween 80 (Sigma-Aldrich), and in ABA treatment, Aquascope (Hellafarm, Greece), were used as wetting agents. No sprayed vines served as control (control treatment). All treatments were applied in triplicate, in sets of 10 vines in a row for each one, in three completely randomized blocks (Supplementary Figure S1a). For the ABA treatment, spraying was performed at the grape zone at veraison stage, and then 3 and 6 days after the first application. In the case of BTH and CHT applications were carried out in the whole vine canopy at veraison and then 7 and 14 days later (Table 1). A total of 200 berries from each treatment were selected randomly. Grape maturity level was monitored weekly by measuring the sugar content, titratable acidity, pH and the weight of 100 berries. Additionally, at the optimum technological maturity stage berries were collected to determine their phenolic maturity level.
For microbial community analyses of the rhizosphere, the phyllosphere and the carposphere, sampling was performed at harvest as follows: from each experimental block of each treatment, four vines were selected, and four soil cores, four samples of leaves and four samples of berries were collected (Supplementary Figure S1b). For ABA-treated vines, since application was performed on the bunches, only berries were sampled for that treatment. The four soil cores were collected from a depth of 0-30 cm and by two were combined, homogenized, sieved (2-mm pore size) and stored at 4 • C until being transferred to the laboratory. For the epiphytic microbial communities of the phyllosphere and the carposphere, the four samples of leaves and berries that were collected were combined into two sterilized falcon tubes and stored at 4 • C until being transferred to the laboratory. A total of 60 samples (18 soil, 18 leaves, 24 fruits) were collected for the analysis of the prokaryotic and fungal microbiome, resulting in six biological replicates per treatment.

Must Conventional Analyses
The main must parameters, sugar content, titratable acidity and pH were determined in accordance with the official methods, as described in the compendium of international methods of wine and must analysis (OIV 2018). In addition, the weight of 100 berries was analyzed (g/berry). The color intensity, tonality, and percentage of blue tones (% blue) were evaluated as indicates in the Glories method (1984). Briefly, 100 grape berries were homogenized using a high-speed Ultra-Turrax at 11,500 rpm for 30 s (T25, IKA, Germany). Fifty grams of homogenized grapes were extracted with (a) 50 mL of 0.1 N HCl solution at pH 1.0 and (b) 50 mL of tartaric acid solution (5 g tartaric acid in 20 mL 1 N NaOH, made up to a 1 L volume with distilled water) at pH 3.2. The samples were shaken at room temperature for 4 h, and then they were centrifuged (13,440× g) at 4 • C for 10 min. The phenolic maturity parameters were determined through a spectrophotometric analysis of the juice. An ultraviolet-visible (UV-Vis) spectrophotometer was used (UV-1900, Shimadzu, Duisburg, Germany). The phenolic richness (A280) was obtained by measuring the absorbance at the wavelength of 280 nm, at pH 3.2, and was expressed as absorbance units (AU) × dilution factor. The anthocyanins extractable at pH 1.0 (ApH1.0) and pH 3.2 (ApH3.2) were measured at 520 nm and expressed as mg of malvidin-3-O-glucoside equivalents/L of extract. The seed maturity index (MP%) was calculated as MP% = [A280 nm − (ApH3.2/1000) × 40]/A280 nm and the skin maturity index was calculated as MS%= 100-MP%

DNA Extraction, Amplicon Sequencing and Bioinformatic Analysis
Soil DNA extraction was performed from 0.25 g of soil with the DNeasy Power soil DNA isolation kit following the manufacturers' protocol (MoBio Laboratories, Carlsbad, CA, USA). Leaves (4-5 leaves) and berries (30-40 g) were washed with a sterilized isotonic solution of phosphate-buffered saline (PBS; 0.1 M) in 50 mL falcon tubes. The epiphytic microbial communities were detached from the surface with four-cycles of 2 min sonication/vortex intervals to achieve maximum recovery of the epiphytic biofilms. Following this, leaves and berries were removed from the tubes with a sterile tong and the solution centrifuged at 8000× g at 4 • C. DNA from the resulting microbial pellets was extracted with the DNeasy Power soil DNA isolation kit following the manufacturer's protocol (MoBio Laboratories, Carlsbad, CA, USA). DNA was quantified using a Qubit 2.0 Fluorometer (Life Technologies, Paisley, UK).
Prokaryotic and fungal communities were analyzed by amplicon sequencing analysis. For prokaryotes, the V4 region of the 16S rRNA gene was amplified with 515f-806r primer pair [40], while for fungi, the ITS2 region was amplified with the primer pair fITS7 [41]-ITS4 [42]. Amplicon multiplexed libraries were constructed with a two-step amplification protocol where a unique 12-bp index was added, either to the forward or reverse primer, to serve as a barcode. Finally, a 2 × 250 bp pair-end sequencing was performed using a HiSeq2500 instrument in Rapid Mode (Illumina, San Diego, CA, USA) at the Genome Sequencing Center of the Brigham Young University (Provo, UT, USA).
The retrieved sequences were demultiplexed with Flexbar v3.0 [43]. In brief, DADA2 [44] was used to remove primers, filter, and denoise sequences, remove chimaeras, merge reads, and construct amplicon sequence variants (ASVs), identify representative sequences of ASVs, and to create an ASV table. The representative sequences of each ASV were classified against the Silva v138 for prokaryotes [45] and UNITE database for fungi [46]. Sequences classified in non-target taxa (e.g., unknown, chloroplasts, mitochondria for the 16S rRNA gene and unknown or protists for the ITS) were removed. Raw sequences were submitted to NCBI under the Bioproject ID PRJNA704932.

Statistical and Diversity Analysis
All analyses were performed with R software [47] and the same steps were followed for both prokaryotic and fungal communities. ASVs, with a relative abundance lower than 0.1% in the two samples, were excluded from downstream analyses. α-diversity indexes for richness (Observed) and diversity (Shannon) were determined with the phyloseq package [48]. The non-parametric Kruskal-Wallis test was used to investigate the main effects of the treatments on α-diversity. β-diversity patterns, thus variations in the structure of microbial communities, were visualized with nonmetric multidimensional scaling (nMDS) based on a Bray-Curtis dissimilarity matrix and permutational multivariate analysis of variance (PERMANOVA; adonis function, with 999 permutations) was applied to assess the effects of the treatments on β-diversity. Significance of differences in the microbial members of the prokaryotic and fungal community in each habitat (rhizosphere, phyllosphere and carposphere) between the treatments (control, CHT, BTH, ABA) were tested using linear discriminant analysis (LDA) and effect size (LEfSe) analysis [49]. Emerged taxa from LefSe analyses was used to generate taxonomic cladograms, illustrating differences between treatments. Finally, Spearman correlation was used to identify significant links between berry chemical characteristics and (i) α-diversity indices per treatment, and (ii) discriminant ASVs from LEfSe analysis. All plots were generated in R with the package ggplot2 [50]. All values are presented as the mean with standard deviation. The significance in the differences in grape berries characteristics was determined with an unpaired t-test or one-way ANOVA with Tukey's test. A p-value ≤ 0.05 was considered to indicate statistical significance. Table 1 shows the influence of the treatments on the standard grape characteristics. A decreasing trend for the mean berry weight after the application of all treatments was observed. The berries treated with CHT, showed a significantly lower value compared to the other treatments. Regarding the total soluble solids (TSS), grapes treated with ABA and BTH showed the highest values. Similar results were recorded for the total acidity (TA), where grapes treated with ABA and BTH showed increased TA. Finally, the values of pH tended to be slightly decreased compared to the control, with a lower value observed for BTH; however, no statistically significant differences were observed among the pH values ( Table 2). Data represents mean ± std. deviation. Different letters in the same column indicate significant differences according to t-test at the 5% probability level.

Phenolic Composition of the Grape Berries
The effects of the treatments on the grape berries phenolic composition using Glories assay are reported in Table 3; Table 4. All elicitors increased the anthocyanins concertation in comparison with the control. The CHT treatment presented the highest values followed by the BTH and ABA treatments. Control grapes had significantly lower anthocyanin concentrations. Extractable anthocyanins (pH 3.2) were also significantly lower in the control grapes. All elicitor/hormone treatments resulted in higher anthocyanin extractability levels, though the difference was not significant for BTH (Table 3). These data indicate that the vine treatments accelerated skin maturation in relationship with the anthocyanin composition. Table 3. Total and extractable anthocyanins of Mouhtaro grapes at harvest stage.

Treatments
Total Anthocyanins (g/L) Extractable Anthocyanins (g/L) Data represents mean ± std. deviation. Different letters in the same column indicate significant differences according to t-test at 5% probability level. Anthocyanin concentration is expressed as g/L of malvidin-3-O-glucoside. Data represents mean ± std. deviation. Different letters in the same column indicate significant differences according to t-test at 5% probability level.
Data related with the tannin's maturity of the grapes are presented in Table 4. The seed maturity index of the CHT-treated grapes (43.2%) was significantly lower compared to all other treatments, including the control (50.5-53.9%). On the contrary, the skin maturity index of the CHT grapes had higher values in comparison with the ABA, BTH, and control grapes. It seems, therefore, that for the CHT-treated berries, there is a higher extraction of tannins from skins than from seeds, in contrast with the other treatments. Hence, the maturity level of the CHT-treated berries is higher regarding the phenolics level, compared to the other treatments, including the control.

α-Diversity Patterns
Initially, quality filtering and ASVs construction, resulted in 12054 ASVs for prokaryotes and 4093 ASVs for fungi. After removal of the ASVs with extremely low relative abundance 1483 prokaryotic ASVs and 537 fungal ASVs, that represent more than 86% of the total communities, remained. α-diversity patterns were plant compartment dependent. For both microbial groups ASV-based richness and Shannon index was higher in the rhizosphere and gradually decreasing in the phyllosphere and the carposphere (Figure 1). The application of the elicitor/hormone treatments significantly affected prokaryote richness and diversity in the rhizosphere, where BTH treated vines showed the lowest index values (p Kruskal-Wallis < 0.001) (Figure 1a). Phyllosphere's richness and diversity were not affected (p Kruskal-Wallis > 0.05), while carposphere's richness and diversity tended to increase in BTH and ABA treated vines, but that increase was not significant (p Kruskal-Wallis > 0.05) (Figure 1a). Rhizospheric fungal community richness and diversity did not change with respect to treatments (p Kruskal-Wallis > 0.05), but differences were detected in phylosphere's diversity and carposphere's richness and diversity ( Figure 1b). BTH treatment increased diversity in the phyllosphere and along with ABA decreased both richness and diversity in the carposphere (p Kruskal-Wallis < 0.01) (Figure 1b).

BTH
50.5 ± 9.6a 49.5 ± 9.6b Data represents mean ± std. deviation. Different letters in the same column indicate significant differences according to t-test at 5% probability level.

α-Diversity Patterns
Initially, quality filtering and ASVs construction, resulted in 12054 ASVs for prokaryotes and 4093 ASVs for fungi. After removal of the ASVs with extremely low relative abundance 1483 prokaryotic ASVs and 537 fungal ASVs, that represent more than 86% of the total communities, remained. α-diversity patterns were plant compartment dependent. For both microbial groups ASV-based richness and Shannon index was higher in the rhizosphere and gradually decreasing in the phyllosphere and the carposphere ( Figure  1). The application of the elicitor/hormone treatments significantly affected prokaryote richness and diversity in the rhizosphere, where BTH treated vines showed the lowest index values (pKruskal-Wallis< 0.001) (Figure 1a). Phyllosphere's richness and diversity were not affected (pKruskal-Wallis > 0.05), while carposphere's richness and diversity tended to increase in BTH and ABA treated vines, but that increase was not significant (pKruskal-Wallis > 0.05) (Figure 1a). Rhizospheric fungal community richness and diversity did not change with respect to treatments (pKruskal-Wallis > 0.05), but differences were detected in phylosphere's diversity and carposphere's richness and diversity (Figure 1b). BTH treatment increased diversity in the phyllosphere and along with ABA decreased both richness and diversity in the carposphere (pKruskal-Wallis < 0.01) (Figure 1b). Based on Spearman's correlation, only non-treated control vines' prokaryotic richness in the carposphere was positively correlated with the o Brix value (Spearman r: 0.77, p < 0.01). For ΑΒΑ, CHT, and BTH there was no link between α-diversity indexes and the physiochemical profile of the berries. Figure 1. The "Observed" (richness) and the "Shannon" (that takes both richness and evenness under account) α-diversity indicators for the ASVs-based proxies of (a) the prokaryotic and (b) the fungal communities of the rhizosphere (soil), phyllosphere (leaf), and carposphere (fruit) of vines under the application, or not (control), of chitosan (CHT), benzothiadiazole (BTH), and abscisic acid (ABA). The upper and lower box boundaries indicate the 75th and the 25th percentiles, respectively; the mid-line indicates the median, the diamond indicates the mean, and the whiskers above and below indicate the 90th and 10th percentiles, respectively; the black dots indicate outliers.

β-Diversity Patterns
Based on Spearman's correlation, only non-treated control vines' prokaryotic richness in the carposphere was positively correlated with the • Brix value (Spearman r: 0.77, p < 0.01). For ABA, CHT, and BTH there was no link between α-diversity indexes and the physiochemical profile of the berries.

β-Diversity Patterns
In terms of β-diversity, the applied treatments were found to affect both prokaryotic and fungal communities differently for rhizosphere, phyllosphere, and carposphere. Bray-Curtis dissimilarity analysis and its visualization through nonmetric multidimensional scaling (NMDS) plots (Figure 2), accompanied by PERMANOVA analysis (Table 5), were adopted to investigate these effects. fungal community variation, while, from the NMDS plots, we see that for both microbial groups that both CHT and BTH did not separate from the control vines (Figure 2c,d). Finally, in the carposphere, we detected a larger effect of treatments, which explained 27.5% of the prokaryotic community variation and 29.5% of the fungal community variation. NMDS plots highlighted that BTH and ABA treatments had a larger effect on prokaryotes compared to control and CHT treated vines, which tended to group together ( Figure 2e) and the fungal community was more distinct for ABA treatment (Figure 2f).

Community Structure and Discriminant ASVs
The composition of the prokaryotic microbial community at the phylum level across the different habitats and treatments is presented in Supplementary Figure S2a and, at the genus level, is presented in Figure 3a. At the phylum level, the rhizosphere was carposphere (e,f). Each dot represents the corresponding community of a single sample (biological replicate). Lines connect the dots with the centroid of each grouping factor, the application or not (control) of chitosan (CTH), benzothiadiazole (BTH), and abscisic acid (ABA). Dashed ellipses represent the 95% confidence interval around the group's centroid. In the rhizosphere, PERMANOVA analysis revealed that applied treatments significantly affected the β-diversity, explaining 26.5% (p < 0.001) and 17.7% (p < 0.001) of the prokaryotic and fungal community variations. NMDS plots demonstrated that, for the prokaryotic community, BTH application resulted in a clearer community separation from the control compared to CHT (Figure 2a), while for fungal communities, treatments were not distinct from the untreated vines (Figure 2b). In the phyllosphere, the applied treatments explained 13.4% of the prokaryotic community variation and 19.3% of the fungal community variation, while, from the NMDS plots, we see that for both microbial groups that both CHT and BTH did not separate from the control vines (Figure 2c,d). Finally, in the carposphere, we detected a larger effect of treatments, which explained 27.5% of the prokaryotic community variation and 29.5% of the fungal community variation. NMDS plots highlighted that BTH and ABA treatments had a larger effect on prokaryotes compared to control and CHT treated vines, which tended to group together ( Figure 2e) and the fungal community was more distinct for ABA treatment (Figure 2f).
At the genus level, sequences were assigned to 424 genera in total. In the rhizosphere, 22 had a relative abundance >1% on average and represented 44.6% of the total community (Figure 3a). The five more dominant genera of the rhizosphere were an unassigned genus of Solirubrobacterales (4.4%), an unassigned genus of Nitrososphaeraceae (3.8), an unassigned genus of Gaiellales (3.8%), Skermanella (3%) and Gaiella (2.3%). Twenty-two genera were found in the phyllosphere with relative abundance >1% on average and represented 56% of the total community (Figure 3a). The five most abundant were Skermanella (12%), Massilia (7.5%), Pseudarthrobacter (3.7%) Blastococcus (3.3%) and Microvirga (3%). In the carposphere, 27 genera showed abundance >1% on average and represented 62% of the total community (Figure 3a) and the five most abundant were Burkholderia-Caballeronia-Paraburkholderia (10.3%), Massilia (5%), Pantoea (4.2%), Cutibacterium (3.3%) and Skermanella (3.2%). Compared to rhizosphere, the genera Lactococcus, Bacillus, and Lactobacillus, were more abundant in both the phyllosphere and the carposphere. Among the treatments, the application of BTH affected the relative abundance of the bacterial genera more in the rhizosphere and the phyllosphere, while, in the carposphere, ABA treatment led to higher abundances of Massilia, an unassigned genus of Planococcaceae and Staphylococcus, and CHT treatment led to higher abundances of Cutibacterium, Lactococcus and Lactobacillus. It is worth mentioning that the genus Oenococcus, members of which are the most important lactic acid bacteria and are most frequently associated with malolactic fermentation in wine, was only present in the carposphere of the ABA-treated berries, but in low relative abundance, 0.08% of the total community. LEfSe results presented as cladograms (Supplementary Figure S3) highlight the significant differences in the microbial members of the prokaryotic and fungal communities in each habitat (rhizosphere, phyllosphere, and carposphere) and between the control and the treatments (CHT, BTH, and ABA). In the rhizosphere, from the 993 prokaryotic and 355 fungal ASVs used in the analyses, only 7% (69 ASVs, 37 for BTH and 16 for CHT) and 4.8% (17 ASVs; 7 for BTH and 10 for CHT) were found to be discriminant for the applied treatments. Discriminant prokaryotic ASVs with a relative abundance more than Fungal sequences were assigned in 233 genera in total. In the rhizosphere, 19 had a relative abundance >1% on average and represented 76% of the total fungal community (Figure 3b). The five most dominant genera of the rhizosphere were Gibberella (12%), Fusarium (11%), Hormonema (8.3%), Solicoccozyma (6%) and an unassigned genus of Agaricomycetes (4.5%). In the phyllosphere, eleven genera had a relative abundance >1% on average and represented 87% of the total community (Figure 3b). The five most abundant were Alternaria (30%), Mycosphaerella (17%), an unassigned genus of Ustilaginaceae (14.5%), Aureobasidium (7.3%) and an unassigned genus of Didymellaceae (6%). In the carposphere, 14 genera had abundance >1% on average and represented 89% of the total community ( Figure 3b) and the five most abundant were Alternaria (23%), Mycosphaerella (17.3%), Cladosporium (11.3%), Aureobasidium (10.5%) and an unassigned genus of Ustilaginaceae (7.5%). The application of the two elicitors and the plant hormone had a lesser effect on the fungal community compared to the prokaryotic community. More specifically, BTH in the rhizosphere increased the relative abundance of Gibberella and Hormonema compared to CHT and the control and in the phyllosphere both BTH and CHT increased the relative abundance of Pseudopithomyces and Epicoccum. In the carposphere among all applications, ABA treatment had a stronger effect by leading to an increased relative abundance of Aureobasidium and a decreased abundance of Cladosporium and Stemphylium. The Saccharomyces genus, members of which are yeasts involved in winemaking, was only present in the carposphere of the ABA-treated berries but in low relative abundance, 0.36% of the total community.
LEfSe results presented as cladograms (Supplementary Figure S3) highlight the significant differences in the microbial members of the prokaryotic and fungal communities in each habitat (rhizosphere, phyllosphere, and carposphere) and between the control and the treatments (CHT, BTH, and ABA). In the rhizosphere, from the 993 prokaryotic and 355 fungal ASVs used in the analyses, only 7% (69 ASVs, 37 for BTH and 16 for CHT) and 4.8% (17 ASVs; 7 for BTH and 10 for CHT) were found to be discriminant for the applied treatments. Discriminant prokaryotic ASVs with a relative abundance more than 1% in BTH treated rhizospheres belonged to genera Candidatus_Nitrososphaera, RB41, Blastococcus, Nitrososphaeraceae_Unknown, Candidatus_Nitrocosmicus, JG30_KF_CM45_Unknown, KD4_98_Unknown, Skermanella and in CHT-treated rhizospheres belonged to Bacillus (Supplementary Table 3). Regarding discriminant fungal members, in the BTH-treated rhizospheres, one ASV had a relative abundance more than 1% and was affiliated to the species Gibberella intricans and in the CHT-treated rhizospheres two ASVs had a relative abundance more than 1% and were affiliated to the species Botryotrichum spirotrichum and Alternaria alternata (Supplementary Table S4).
Treatments had an even milder effect on the microbial communities in the phyllosphere, where from 1015 prokaryotic and 282 fungal ASVs used in the analyses, only 2.4% (24 ASVs, 13 for BTH and 11 for CHT) and 2.5% (7 ASVs, 5 for BTH and 2 for CHT) were found to be discriminant for the applied treatments. Discriminant prokaryotic ASVs with relative abundance more than 1% in BTH-treated phyllospheres belonged to the genera Massilia and Pseudarthrobacter and in CHT-treated phyllospheres belonged to Skermanella (Supplementary Table S5). Regarding discriminant fungal members, in BTH-treated phyllospheres, one ASV, affiliated to Pseudopithomyces rosae, had a relative abundance more than 1% and one in CHT-treated phyllospheres affiliated to the species Epicoccum_Unknown (Supplementary Table S6).

Linking Members of the Microbial Communitites in the Carposhere with Berries Chemical Characteristics
In a following step, correlation of the prokaryotic and fungal ASVs of the carposphere, which emerged from the LefSe analyses as a discriminant for the three treatments, with the chemical characteristics of berries was assessed (Figure 4). It should be mentioned that the observed associations indicate linkages rather than causality. Each treatment showed a unique pattern in the characteristics that the ASVs are correlated with, and this pattern is consistent across both prokaryotes and fungi. The majority of the ABAdiscriminant prokaryotic and fungal ASVs had positive correlations with • Brix degrees. Several prokaryotic ASVs correlated positively with the seed and cellular maturity index and with extractable anthocyanins, while two showed negative associations with total acidity. Similarly, fungal ASVs, which are mainly yeasts, showed a positive association with extractable anthocyanins and a negative association with total acidity. Concerning BTH treatment, only the prokaryotic ASVs were associated with the berries' chemical characteristics. These ASVs showed a positive correlation with the weight of the berry, • Brix degrees, and the seed maturity index, while they demonstrated negative correlations with total acidity, extractable and total anthocyanins, and the cellular maturity index. Finally, regarding CHT treatment, the emerged ASVs from both prokatyotic and fungal community showed the opposite patterns. Prokaryotic ASVs included some lactic acid bacterial genera, showed negative associations with the weight of the berry, • Brix degrees, and the seed maturity index but presented positive associations with total acidity, extractable and total anthocyanins, and the cellular maturity index. A similar pattern was recorded for fungal ASVs, as they showed negative associations with • Brix degrees, pH and berry weight and positive associations with total acidity.

Discussion
Exogenous application of molecules that elicit plant defense mechanisms have been thus far studied as an alternative to conventional pathogen management [51]. However, there is a growing body of literature that investigates the effects of elicitors and plant hormones on grape and wine quality characteristics and, more specifically, on potential improvements to the biosynthesis of phenolic compounds in grapes, and subsequently in produced wines [5,6,23,52]. From a nutritional point of view, grape products are functional foods that provide health benefits beyond basic nutrition, by virtue of their content of nutraceuticals and bioactive compounds, such as polyphenols, and their pharmaconutritional properties [53].
In the present study two elicitors and a major plant hormone were applied in a single vineyard of Mouhtaro, a wine grape variety commonly used in the production of a monovarietal wine, in Viotia, and more especially in the Muses Valley. Our main purpose was to determine the effect of these compounds on (i) the characteristics involved in the technological and phenolic maturity of grapes during the harvest stage, and (ii) the microbial ecology in the rhizosphere and on the surface of leaves and berries, which may also be directly or indirectly affected via metabolic and physiological plant responses. To our knowledge, this is the first report investigating the effect of two resistance elicitors, chitosan and benzothiadiazole, and a plant hormone, abscisic acid, on grape must standard characteristics and indices, which define the so-called technological maturity of the grapes, as well as on the phenolic composition of the grape berries and on the microbial communities in close association with different plant compartment.
One of the most crucial phenotypic stages in grapevine is veraison. During this stage, berries start gaining color due to anthocyanins and other phenolic compounds that accumulate in berry skins. The application of CHT, ABA, and BTH increased the synthesis of anthocyanins and other phenolics in berries compared to the untreated control berries, highlighting their potential for eliciting the biosynthesis of these secondary metabolites' pathways during veraison stage. Results are in accordance with previous studies that showed increased values of anthocyanins and phenolic compounds in grape berries [12,54,55]. It is also important for wine production, as wines produced from ripper grapes had a more intense red color and were also less astringent due to the lesser contribution on the tannins of the seeds, which are more astringent due to the higher galloytation content [50]. There is evidence that chitosan treatment may activate key enzymes of the phenylpropanoid pathway, in particular phenylalanine ammonia lyase, which is the key enzyme that catalyzes the first step in phenolic biosynthesis [51].
The chemical analysis of the grape musts at the harvest stage did show significant differences (weight/berry, TSS and TA) related to the various treatments applied ( Table 2). Romanazzi et al. [56] showed that grape musts from Montepulciano grapevines treated with laminarin and copper hydroxide were not significantly different from controls, while Rantsiou et al. [57] recorded similar results after the application of fungicides on the Nebbiolo variety. However, significant differences were highlighted in the total phenolics and anthocyanins in grape berries after the application of the treatments. The highest values of total anthocyanins (Table 3) and the skin maturity index (Table 4) were recorded for the CHT-treated grape berries among the grapes. This is an advantage from an oenological point of view, since the grape skins are the major source of color and aroma compounds [58]. However, the CHT-treated berries showed the lowest values in the seed maturity index ( Table 4).
The goal was to better understand the induction and biochemical synthesis of grape polyphenolic compounds, and to improve viticultural techniques, both of which will prove very useful in the field of plant protection, such as conferring resistance in pathogenic micro-organisms, as well as obtaining the optimum organoleptic properties.
The role of microbial consortia residing inside and outside plant tissues, or even in the rhizosphere, should be considered as another important factor in the development of compounds important for grapevine protection and, potentially, for wine quality characteristics [59,60]. From an enological perspective, however, limited knowledge is available on how to practically manipulate microbial plant communities on/in the grapevine to improve vine/grape characteristics.
Even though these kinds of effects are of significant importance, they have still not been deeply investigated. Next generation sequencing and metagenomics tools are essential contributors in clarifying if there are differences in the microbial communities induced by the application of elicitors, and also if these communities are hidden producers of metabolites, including polyphenolic compounds. Due to the multiple microbial communities in the grapevine, however, it is difficult to determine which microorganisms can contribute to the biosynthesis of phenolic compounds.
Plants are a habitat for diverse microbial communities. The members of these communities interact with each other and with their host plant, and this communication is triggered by molecule and signals and, at the same time, triggers the biosynthesis of other molecules [61]. Similarly, biotic (e.g., plant pathogens or pests) and abiotic (e.g., exogenous application of elicitors or biostimulants) factors have direct or indirect effects (through changes in plant physiology) on the plant-associated microbial communities. In the present study, we examined both prokaryotic and fungal microbial communities in the rhizosphere, phyllosphere and carposphere of grapevines, which received a foliar spray of two elicitors, CHT and BTH, and a berry spray of a plant hormone, ABA. Among the different plant compartments, both nMDS plots ( Figure 2) and PERMANOVA analysis (Table 5) highlighted that the prokaryotes of rhizospheric communities were mostly affected, followed by those of the carposphere, while for fungal communities, mainly those of the carposphere communities were affected. In general, fungal communities were found to be less sensitive to treatments compare to prokaryotic communities.
Concerning the effects of BTH and CHT on the microbial communities of the rhizosphere, this could reflect a systemic mode of action of these plant activators. Both molecules trigger either of the two main plant defense pathways, SAR and ISR. The activation of these pathways alters the chemical composition of plant root exudates [62], which in turn could lead to modifications to the rhizospheric microbial community [59,62]. Interestingly, even though CHT and BTH was sprayed on leaves, the microbial communities of the phyllosphere were less affected (Table 5). Likewise, phyllosphere microbial communities were slightly modified after foliar application of a protein-based resistance inducer of greenhouse-grown grapevines [36]. Leaves' surfaces are subjected to continuous microbial "inoculation" from aerial transportation [62], this, combined with the long period between treatment application and harvest, gave time to the phyllosphere microbial communities to either return to their original or an alternative state, masking the treatment effects.
In the carposphere, treatments highly affected both prokaryotic and fungal communities. The treatment-discriminated microbes appeared to be zymogenous strategists (boosted by sugars and a higher pH) in the case of ABA and BTH, but were also autochtonous strategists (boosted by phenolics and acidity) in the case of CHT. More specifically, ABA treatment increased the relative abundance of native yeasts and decreased those of putative fungal pathogens, while CHT-treated grapes demonstrated an increase in lactic acid bacteria. Both microbial groups are of great importance for the winemaking process. The use of native yeasts could be considered a potential strategy for controlling the presence of mycotoxinproducing molds. According to the results of the present study, ABA treatment could increase the relative abundance of beneficial yeast on grapes and, at the same time, decrease the percentage of pathogenic fungi. Debaryomyces hansenii is the most famous species of the genus Debaryomyces, and is known for its antimicrobial properties towards undesirable species [63]. For instance, D. hansenii acts against the plant pathogen Aspergillus parasiticus through repression of regulatory genes implicated in aflatoxin biosynthesis, but other modes of action have been also mentioned, such as the competition for space and nutrients, as well as the production of enzymes, such as proteases and chitinases [64].
Lactic acid bacteria have also been used as biocontrol agents in winemaking [65]. The capacity of LAB to produce exopolysaccharides and form biofilms on different oenological surfaces help them to persist during the various steps of winemaking, from the grape berries to alcoholic and malolactic fermentation, up to the bottled wine [66]. Even if Oenococcus is the most widely used bacterial genus in winemaking [67], Lactobacillus strains have started to gain increasing interest in the wine industry due to their promising properties as malolactic starters [68]. Lucio et al. [69] tested different strains from various Lactobacillus species in order to drive malolactic fermentation in grape must. Based on their results, the tested strains of Lactiplantibacillus plantarum expressed the best overall performance in grape must and were, not only able to deplete malic acid, but also contributed to wine flavor. Interestingly, in our study we showed that CHT treatment could increase the abundance of members of the Lactobacillus genus on grapes compared to the control, as well as to the other tested treatments. It would be interesting in the future to examine the oenological properties of these bacteria to propose an intergraded biocontrol solution.
In addition, both prokaryotic and fungal ASVs that emerged from LefSe analyses as discriminants demonstrated the same correlation patterns at berries characteristics with respect to the ABA and CHT treatments, while BTH only had a significant effect on prokaryotic microbial communities (Figure 4). ABA discriminant prokaryotic and fungal ASVs correlated positively with berries' physicochemical characteristics and the seed maturity index, while CHT-discriminant prokaryotic and fungal ASVs correlated negatively. The opposite pattern was observed for anthocyanins and the skin maturity index. These patterns could imply that discriminant ASVs probably act in a similar way or share a common functional traits.

Conclusions
The application of plant elicitors and hormones as a viticulture practice represents a sustainable way to reduce fertilizers and other chemicals, leading to better management of environmental contamination. Our results indicate that the use of plant elicitors and hormones deserves particular attention, not only because of their low environmental impact, but also for their implications in the improvement of quality traits. Their exogenous application during the veraison stage significantly changed grape chemical and phenolic parameters. Therefore, such plant treatments can be used in the field of viticulture in the new era as viticultural techniques to improve quality characteristics of berries and of produced wines. In addition, the application of these compounds may influence microbial communities, which reside on the surface of berries and leaves, as well as in the rhizosphere soil, which influences the health and metabolic responses of grapevines. As a consequence, the model of grapevine, microorganisms, quality characteristics of grape berries and plant activators could be seen in holistic way, where all these factors could all interact together in order to contribute to the sustainability of a vineyard. Due to the potential biotechnological application of microbes in agriculture, studies to link the microbial compositions of soil, grape and leaves and the wines features themselves have intensified. As this study indicates, plant activators should not be considered as a one-size-fits-all option and the significant challenge will be had in determining the best variety × environment × management practices (including plant activators) in order to select the best combinations.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/su13115802/s1, Figure S1: (a) Prokaryotic and (b) fungal community composition at the phylum level (ten most abundant phyla shown) across habitats (rhizosphere: soil; phyllosphere: leaf and carposphere: fruit) and treatments (control, CHT, BTH, ABA). Significant differences between treatments for each habitat are shown in Supplementary Table S1. Figure S2: Cladograms based on linear discriminant analysis (LDA) effect size (LEfSe) analysis of the (a,c,e) prokaryotic and (b,d,f) fungal communities among all sampling sites in the rhizosphere (a,b), phyllosphere (c,d) and carposphere (e,f). Table S1: The prokaryotic community composition at the phylum level (ten most abundant phyla shown) across habitats and treatments (control, CHT, BTH, ABA). Significant differences between treatments within each habitat and for each phylum are indicated with different letters at p < 0.05 (Tukey's post hoc test). Table S2: The fungal community composition at the phylum level (ten most abundant phyla shown) across habitats and treatments (control, CHT, BTH, ABA). Significant differences between treatments within each habitat and for each phylum are indicated with different letters at p < 0.05 (Tukey's post hoc test). Table S3: The prokaryotic community members in the rhizosphere that emerged from LefSe analysis as discriminant for BTH and CHT treatments. Table S4: The fungal community members in the rhizosphere that emerged from LefSe analysis as discriminant for BTH and CHT treatments. Table S5: The prokaryotic community members in the phyllosphere that emerged from LefSe analyses as discriminant for BTH and CHT treatments, Table S6: The fungal community members in the phyllosphere that emerged from LefSe analysis as discriminant for BTH and CHT treatments. Table S7: The prokaryotic community members in the carposphere that emerged from LefSe analysis as discriminant for ABA, BTH and CHT treatments. Table S8: The fungal community members in the carposphere that emerged from LefSe analyses as discriminant for BTH and CHT treatments.