Mould incidence was recorded as the percentage of mouldy berries per replicate, according to Youssef et al. (2019). Afterwards, grape skin colour was determined using a hand-held tristimulus colourimeter with a D65 illuminant and 8 mm light aperture (model CR-400 Chroma Meter, Konica Minolta Inc., Cheshire, UK). The colourimeter was calibrated with a white plate, and it provided the lightness (L*), chroma index, and Hue angle values of the samples. These parameters are widely used as a measure of ripeness (Conesa et al., 2016).

The firmness of individual berries was determined using a uniaxial testing machine calibrated with a 5 N cell (Instron, model 5542, MA, USA). Maximum compressive load in Newtons (N) was measured by compressing the berry 1 mm deep with a 4 mm diameter probe at a speed of 100 mm min^-1 (Alamar et al., 2017).

Respiration Rate (RR) was measured as described in Collings et al. (2018). Samples were placed in 3 L airtight glass jars and dry clean air (300 mL min^-1) was passed through the jars into a Sable Respirometry System (model 1.3.8 Pro, Sable Systems International, Las Vegas, USA). Each sample jar (n = 3 replicates per treatment) was measured for 2 min to achieve a stable reading. An empty jar (continuous air for 1 min) was used as a baseline to avoid cross-contamination between treatment measurements. The rate of CO₂ production was expressed in mL kg^-1 h^-1.

Extraction and analysis of individual sugars (glucose, fructose, and sucrose) were done following Davis et al. (2007). Briefly, 150 mg of freeze-dried fruit samples were extracted with 3 mL of 62.5% (v/v) aqueous High Performance Liquid Chromatography (HPLC)-grade methanol in a water bath at 55°C for 15 min, vortexing every 5 min for 30 s. The extract was then filtered through 0.2 μm PTFE filters and stored in 2 mL HPLC vials at -40°C until analysis. Before analysis, the extracts were diluted 1:9 with HPLC grade water.

Detection was performed using a Refractive Index Detector (RID) coupled to an Agilent 1200 series HPLC system (Agilent Technologies, Germany) with a Phenomenex Rezex RCM monosaccharide Ca⁺² (8%) ion exclusion column (300 mm x 7.8 mm), 8 μm particle size, fitted with a Phenomenex Carbo Ca⁺⁴ mm x 3 mm guard column (Phenomenex, CA). The column oven temperature was set at 80°C and the refractive index detector at 50°C. The mobile phase used was HPLC-grade water at a flow rate of 0.6 mL min^-1, and the autosampler injection volume was 20 μL. A 6-point calibration curve ranging from 0.05 to 2.5 mg mL^-1 was used for quantification, and the results were expressed as mg g^-1 dry weight (DW).

Analysis of non-volatile organic acids was done as described in Terry et al. (2007) with slight modifications. Briefly, freeze-dried samples were dissolved in 3% aqueous metaphosphoric acid, kept at room temperature for 10 min, filtered through 0.2 μm cellulose filters into HPLC vials, and analysed immediately after extraction.

The acids content was determined using an Agilent 1200 series HPLC system with an Alltech Prevail Organic Acid 250 mm x 4.6 mm, 5 μm particle size column fitted with an Alltech Prevail Organic Acid 7.5 mm x 4.6 mm, 5 μm particle size guard column (Alltech, CA, USA). The column oven temperature was set at 35°C, and the compounds were detected by a diode array detector (DAD) set at 210 nm. The mobile phase used was 25 mM KH₂PO₄ HPLC-grade water at a flow rate of 1.5 mL min^-1, and the autosampler injection volume was 20 μL. A 5-point calibration curve ranging from 0.05 to 1 mg mL^-1 with a mixture of tartaric and malic acids was used for quantification, and the results were expressed as mg g^-1 DW.

ABA and ABA catabolites concentration was determined by following Serradilla et al. (2019) with modifications. Freeze-dried samples (5 mg) were mixed with 500 μL of methanol/formic acid/water (60:5:35, v/v) and 10 μL of the internal standards mix (100 ng mL^-1) in a 1.5 mL Eppendorf tube, and shaken in a Star Beater (R&L Slaughter Ltd., Essex, UK) at 30 Hz for 2 min. The internal standard mix was made of the labelled forms of the compounds: (-)-5,8'8'8'-d4-abscisic acid (d4-ABA); (-)-7',7',7'-d3-phaseic acid (d3-PA); (-)-7',7',7'-d3-dihydrophaseic acid (d3-DPA); (+)-4,5,8',8',8'd5-abscisic acid glucose ester (d5-ABA-GE); (±)-5,8',8',8'-d4-7'-dihydroxy-ABA (d4−7 -OH-ABA). The tubes were then placed in dry ice and kept in darkness for 20 min. Afterwards, the tubes were centrifuged at 4°C and 14,000 rpm for 10 min, then transferred to 15 mL falcon tubes and freeze-dried overnight. The extracts were reconstituted with 500 μL of acetonitrile/formic acid/water (10:0.1:89.9, v/v), vortexed for 1 min, sonicated for 1 min, and centrifuged at 4°C and 4,500 rpm for 1 min. The samples (20 μL) were then injected on a Phenomenex Luna C18 100 mm x 2 mm, 3 μm with guard column at 40 °C, and analysed by an Agilent 1200 series HPLC system coupled to a Q-Trap 6500 mass spectrometer (AB Sciex, MA, USA). The concentration of ABA and ABA metabolites was quantified based on Morris et al. (2018) methodology. A 6-point calibration curve ranging from 0.1 to 100 ng mL^-1 was used for quantification, and the results were expressed as ng g^-1 DW.

ABA is known to regulate plant growth and developmental processes, such as cell division, seed dormancy, maturation, ripening, and abscission, as well as stress responses to drought, cold, or pathogen attack (Finkelstein, 2013; Jia et al., 2017; Coelho et al., 2019). While a few studies have investigated the profile and role of ABA and its catabolites during the preharvest stages of grape berry development and ripening, less attention has been given to the hormonal profile of berries during postharvest and senescence. According to Coombe and Hale (1973), the onset and rate of berry ripening are a function of ABA accumulation. The authors suggested that berry ripening starts once ABA accumulates to a certain level (viz. 62 ng g^-1 fresh weight [FW]), below which the berry will not ripen. In the literature, there is a considerable variation in the value of this peak and the moment in berry development when it appears. For instance, Coombe and Hale (1973) observed that ‘Doradillo’ grape veraison started with an ABA concentration of 62 ng g^-1 FW, which then increased to 212 ng g^-1 FW 13 days after veraison. In contrast, Sun et al. (2010) reported an ABA peak of 500 ng g^-1 FW two days after ‘Muscat Hamburg’ grape veraison and a second ABA peak at similar levels six days after harvest after which the berries started to senesce. In a study by Owen et al. (2009), the ABA concentration in the pulp of immature ‘Merlot’ grape berries was 4,000 ng g^-1 DW, while mature berries harvested at commercial maturity had an ABA content of 400 ng g^-1 DW. Moreover, the authors found that ABA was catabolised to form ABA-GE by conjugation instead of producing DPA by oxidation at the onset of veraison. However, the mentioned study did not quantify ABA or its catabolites during the postharvest berry senescence.

In this work, ABA accumulation coincided with a decrease in the organic acid content and firmness of the berries and an increase in berry respiration rate, sugar hydrolysis, and disease incidence, which are factors determining quality acceptance for the final consumer (Wolf, 2002; Jianying et al., 2014). These results suggest that the grape berries started to senesce after reaching an ABA concentration higher than 100 ng g^-1 DW and that ABA played a role in regulating grape berry metabolism during storage and shelf-life. The crosstalk between ABA and sucrose in regulating fruit ripening processes has been recently reviewed (Durán-Soria et al., 2020; Alferez et al., 2021). Jia et al. (2017) found that treatments with exogenous ABA and sucrose regulate berry ripening of ‘Fujiminori’, a black table grape, and that sucrose can induce the ABA synthesis gene expression. According to Luo et al. (2020), ABA and sucrose play a synergistic role in regulating the ripening of strawberry, a non-climacteric produce. Our results were not conclusive in this area; even though we found a negative correlation between sucrose and ABA, there was not enough evidence to consider this correlation as proof for signalling the triggering of senescence processes.

In addition to regulating sugar levels, ABA is involved in the biosynthesis of secondary metabolites like anthocyanins, which are pigments that give red, purple, and blue colours to plants (Owen et al., 2009; Zhang et al., 2014). For this reason, ABA is applied to coloured grape varieties during berry maturation to improve skin colour development by enhancing the accumulation of anthocyanin biosynthetic enzyme genes (Yamamoto et al., 2015; Domingues Neto et al., 2017). In this study, an increase in the fructose and glucose content coincided with darker berry colour and a higher ABA content by the end of storage and shelf-life, confirming the role of sugars in the stimulation of pigmentation (Durán-Soria et al., 2020). Darker berry colour is related to higher consumer acceptability, and therefore, the understanding of ABA metabolism and its interaction with pigment development is key for the grape industry.

ABA-GE concentration was ten times higher than ABA, playing an important role in grape berry ripening, especially on stressed fruit (Zarrouk et al., 2016). The highest peak in ABA-GE accumulation occurred for berries subjected to 1-MCP at 0.5 C with the jump of temperature from storage to shelf life stages, showing an abiotic stress as an adjustment to the changing environment. The general trend for ABA-GE concentration was increasing, linking to berry development (Zarrouk et al., 2016). It has been demonstrated that ABA-GE can regulate ABA accumulation; this is interesting because the increase of ABA concentrations follows a previous increase in ABA-GE concentrations, suggesting that ABA-GE can play a role in ABA synthesis through β-glucosidase in agreement with e.g., López-Carbonell et al. (2009) and Xu et al. (2014).

ABA is stored in the form of ABA-GE in the vacuoles and the apoplast during the early berry developmental stages, being released to the endoplasmic reticulum under water stress or dehydration (Finkelstein, 2013; Jia et al., 2017). Thus, it could be hypothesised that the proximal section would have a higher hormonal content due to dehydration in the abscission zone, as ripening related changes in the cell wall undergo from the formation of a dehiscence zone to the softening of the flesh tissue (McAtee et al., 2013). Contrastingly, here it was found that the concentration of ABA and its metabolites, mainly ABA-GE, was significantly higher (3-fold) in the distal section when compared to the proximal section, confirming the hypothesis of a different hormonal distribution within the grape berry during senescence. Two facts could explain this: firstly, the distal section of the berries had a darker colour than the proximal, and secondly, it was observed that this section also presented a higher mould incidence during the last stages of cold storage and shelf-life.

Conversely, DPA concentration was ten times lower than ABA in agreement with a previous study by García-Pastor et al. (2021a) on ‘Magenta’ table grape. DPA showed an increasing trend, particularly during shelf life as seen in Pogány et al. (2020).

High levels of DPA have been observed during the early developmental stages (Owen et al., 2009). Therefore, a low level of this hormone could potentially indicate an advanced level of ripening or senescence. A larger sample size and a deeper dive in molecular biology pathways could help identify the involvement of DPA in ripening and stress responses ( Supplementary Figure 2 ).