Effects of extra potassium supply and rootstocks indicate links between water, solutes and energy in Shiraz grapevines (Vitis vinifera) pericarps

K fertiliser trial. Impact of increased K supply rate during berry ripening in potted Shiraz vines

Rootstock trial. Comparison of berry K levels in Shiraz grown on own rooted vines or 420-A, Ramsey or Ruggeri-140 rootstocks

Shiraz grapevine berries (Clone BVCR12), located at the Waite Campus of the University of Adelaide, SA, Australia, were sampled from vines on their own roots or one of the three rootstocks: (1) Ramsey; (2) Ruggeri-140; and (3) 420-A. Berry samples were collected from the eastern and western sides of the north-south oriented rows of the canopy in the 2018/2019 and the 2019/2020 seasons. Ten replicates, each containing 11–26 berries, were sampled weekly from veraison (65 DAA in the 2018/2019 season; 73 DAA in the 2019/2020 season) until the end of ripening (108 DAA in the 2018/2019 season; 114 DAA in the 2019/2020 season). During sampling, berries were snap frozen in LN₂ in the vineyard, and then stored at −80°C until processing and analysis. The frozen ground berry samples with recorded fresh mass (470–530 mg) were dried at 65°C for 3 days. Dried berry samples were extracted by 2 mL nitric acid (0.22 M) in 50 mL cap-sealed tubes (352070, Falcon; 430829, Corning), which was incubated on a hot block (SCP Science, DigiPREP) at 100°C for 2 h. After extraction, solids were removed by centrifugation at 15,000g for 10 min. The supernatants were collected as K extracts. After 10-fold dilution with MilliQ water, the K concentration of extracts was measured by flame photometry (Model 420 Flame Photometer, Sherwood Scientific) following the manufacturer’s instructions. The results from flame photometry, K extract volume, dilution factor and sample fresh mass were used to calculate the K concentration and content of the berry sample.

Data visualisation and statistical analysis

Data were processed and visualised using Excel and MATLAB. N-way ANOVA with two-way interactions and type three sums of squares, multiple comparisons following ANOVA by Tukey’s honestly significant difference criterion, correlation coefficient, linear regression, and principal component analysis (PCA) were undertaken using MATLAB ver. 9.13.0.2105380 (R2022b). Two-way ANOVA was conducted on inorganic elements in vegetative tissue, to assess the effect of tissue and fertiliser K level. Two-way repeated measures ANOVA were undertaken on pericarp parameters in both trials. The ANOVA conducted for the K fertiliser trial involved two explanatory factors, DAA and fertiliser K level as fixed effects, as well as one within-subject factor, plant, as a random factor. The ANOVA conducted for the rootstock trial examined two explanatory factors, DAA and rootstock as fixed effects, and the within-subject factor, plant, as a random factor. PCA was undertaken using mean centred and variance-scaled data block in singular value decomposition. A two-way repeated measures ANOVA with explanatory factors DAA and rootstock as well as vine as the subject factors was undertaken on the concentration and content of berry K in 2018/2019 season and 2019/2020 season.

The significance of the factor contribution to data variation was indicated by the p value of the ANOVA (Tables S1–S3). Assessed by multiple comparison tests following ANOVA, significant differences (P < 0.05) between experimental groups on the same sampling date are shown by asterisks in the figures, and by the P-values in the ‘Results’ and ‘Discussion’ sections.

In the K fertiliser trial, the box plot, showing the median, the lower and upper quartiles, any outliers (dots in the figures), and the minimum and maximum values other than outliers, were used to demonstrate the distribution of sample data. The box plots for the two treatment groups were plotted in the same figures to demonstrate the alteration in berry components on the same sampling date, as well as in developmental evolution patterns across the berry ripening period. Boxes of two groups on the same sampling date were separated, resulting in 14 boxes on 7 sampling dates. In the rootstock trial, the bar chart along with the error bar representing the standard error of the mean was used to show the impact of rootstock on every sampling date.

Doubled-K fertilisation increased pericarp L-malic acid content in the earlier phase of ripening

Shiraz vines allocated to the experimental group were supplied with a doubled K rate compared with those in the control group supplied with a standard K rate. The elements Ca, K, Mg, Na and P, as well as the ratio of K to Na, were analysed in the petiole, rachis and pedicel to assess the vine nutrient status (Fig. S2). Two-way ANOVA for the inorganic elements concentrations indicated that only the uptake of K into the experimental vines was significantly increased in the Doubled-K fertiliser group (P = 0.0374) (Table S1). The K level in the fertiliser did not affect the concentrations of other elements or the tissue K/Na ratio. Element concentrations in the same tissue type did not differ under the various fertilisers, as indicated by the multiple comparisons test following ANOVA (Table S1). The concentrations of the inorganic elements in the vegetative tissues were indicators of their availability for transport into berries.

Shiraz pericarp parameters, including level of hydration, metabolites and inorganic elements, followed the same trends for vines under Doubled-K fertiliser as those under Standard-K supply (Figs 1, S3 and S4). In the Standard-K fertilisation, L-malic acid concentration declined exponentially from 140 μmol g⁻¹ pericarp fresh mass at 50 DAA to 20 μmol g⁻¹ pericarp fresh mass at 103 DAA with further minor decreases in later ripening, attaining 10 μmol g⁻¹ pericarp fresh mass (Fig. 1a). The Doubled-K fertiliser resulted in similar trends, with final L-malic acid concentrations at 13 μmol g⁻¹ pericarp fresh mass at 143 DAA. In addition, pericarp L-malic acid content in both groups declined linearly at a relatively consistent rate across ripening, except for the middle phase of ripening (77 DAA) when the content rose in both treatments (Fig. 1b). Doubling K supply contributed to the variation in pericarp L-malic acid content via interaction with DAA (P = 0.0027) (Table S2), significantly increasing the value at 77 DAA compared with Standard-K supply (P = 0.00008) as indicated by the multiple comparisons test following ANOVA. The pericarp L-malic acid concentration during berry ripening varied significantly between biological replicates, indicated by the significant interaction between DAA and plant (P = 0.005) (Table S2).

Fig. 1.

Seasonal evolution of pericarp concentration and content of L-malic acid (a, b), K (c, d), Mg (e, f), and Na (g, h), as well as the pericarp K/Na ratio (i) compared with those of vegetative tissues (j) in the Shiraz K fertiliser trial. (a–i) Vertical dash lines represent the onset of ripening (veraison). Box plots from two groups are separated by sampling dates 50, 63, 77, 88, 103, 123, or 143 DAA. *P < 0.05 between the two treatments on the same sampling date; ****P < 0.0001 between the two treatments on 77 DAA. (j) Mean values are presented with error bars indicating the s.e.m. n = 7.

Shiraz pericarp fresh mass increased more than two-fold (P < 0.0001) between 63 and 77 DAA to peak at 2.5 g and then stabilised until 103 DAA prior to declining by 25% at 143 DAA (P < 0.001) (Fig. S3a). There was no impact of fertiliser on pericarp fresh mass (P = 0.9558). As expected, pericarp water mass (Fig. S3c) followed similar trends as fresh mass. Pericarp dry mass did not decline during the later ripening phase (Fig. S3d). TSS increased linearly across the sampling period in both the Standard-K fertiliser (r² = 0.904, P < 0.0001). and Doubled-K fertiliser group (r² = 0.918, P < 0.0001) from five to 30°Brix (Fig. S3e).

Pericarp total sugar concentration also changed over time, increasing more rapidly during the earlier ripening phase, relative to the later phase (Fig. S4a) and peaking at 1 mmol g⁻¹ pericarp fresh mass. Pericarp total sugar content ceased to increase beyond 103 DAA, peaking at 2 mmol per pericarp (Fig. S4b), with similar accumulation patterns as that of water. No fertiliser effects were apparent on sugar concentration (P = 0.6493) or content (P = 0.7006) as assessed by ANOVA with no main treatment, but there were significant interactive effects between the fertiliser and biological (plant to plant) variation on the sugar concentration (P = 0.0001) and sugar content (P = 0.0006) (Table S2).

Pericarp ATP concentration and content were not altered by K fertiliser rates (Fig. S4c, d), although biological (plant to plant) variation significantly impacted ATP concentration (P = 0.0357) (Table S2). In both treatments, pericarp ATP concentration was highest at the earliest sampling date with 0.11 μmol g⁻¹ pericarp fresh mass and subsequently declined 75% to 0.03 μmol g⁻¹ at the last sampling date (P < 0.0001, Fig. S4c). Overall, ATP concentration was between 1% and 14% lower in the Doubled-K fertiliser treatment than in the Standard-K supply (P = 0.0491). ATP content followed a different trend, with a maximum value at 123 DAA prior to declining 60% at 143 DAA (P < 0.0001, Fig. S4d). Pericarp ATP content in the Standard-K fertiliser treatment was higher than those of the Doubled-K fertiliser group (P = 0.0226) by 6% to 15% across berry development.

The Doubled-K supply did not alter pericarp ethanol concentration and content. Ethanol concentration steadily increased during berry ripening, peaking at about 3.5 μmol g⁻¹ pericarp fresh mass at 143 DAA (Fig. S4e). The pericarp ethanol content displayed a similar trend to that in pericarp ATP content, with the maximum values detected at 123 DAA (Fig. S4f).

Doubled-K fertilisation raised the K and Mg concentrations and reduced K/Na ratio in mature pericarps

There were significant two-way interactions between the effects of berry development (DAA) and fertiliser on pericarp K concentration (P = 0.0092) and K content (P = 0.0161) (Table S2). K concentration continuously increased during berry ripening, especially between 77 DAA and 88 DAA (P < 0.01), and between 123 DAA and 143 DAA (P < 0.0001) (Fig. 1c). Doubled-K fertiliser significantly increased pericarp K concentration at 103 DAA (P = 0.0478) and 123 DAA (P = 0.0188), compared with the Standard-K treatment on the same sampling dates. The final pericarp K concentrations were 128 μmol g⁻¹ pericarp fresh mass in the Standard-K fertiliser and 138 μmol g⁻¹ pericarp fresh mass in the Doubled-K fertiliser treatment. Pericarp K content increased three-fold between 63 DAA (veraison) and 77 DAA (P < 0.0001) in the same treatment group (Fig. 1d). Pericarp K content peaked at 88 DAA in the Standard-K fertiliser group with 190 μmol per pericarp, however pericarp K content in the Doubled-K fertiliser group further increased to 220 μmol per pericarp at 103 DAA and 123 DAA. Pericarps of vines supplied with Doubled-K fertiliser contained significantly higher K content at 103 DAA (P = 0.0159) and 123 DAA (P = 0.0188) than those in the Standard-K supply group.

K content did not differ between Standard- and Doubled-K fertiliser groups at 143 DAA (P = 0.6046). Conversely, pericarp Mg accumulated slowly in the earlier ripening phase until after 103 DAA, reaching the final concentration between 4.5 and 7 μmol g⁻¹ pericarp fresh mass in both treatment groups (Fig. 1e). Compared with the Standard-K treatment, Doubled-K fertiliser increased the Mg concentration significantly only at 123 DAA (P = 0.0451) than the Standard-K group on this date. The effect of fertiliser on pericarp Mg concentration interacted significantly with the effect of DAA (P = 0.0103) (Table S2). Maximum pericarp Mg content was delayed from 88 DAA in the Standard-K group to 103 DAA in the Doubled-K fertiliser group, with the maximal values of 8 or 9 μmol per pericarp, respectively (Fig. 1f). Pericarp Na concentration and content were higher in the Doubled-K fertiliser group, but the variation between two treatments was not significant (P > 0.05), which was associated with the more variable data in the Doubled-K fertiliser group Fig. 1g, h).

Pericarp P concentration increased steadily during berry ripening (Fig. S4g), following a similar pattern to that of pericarp K concentration. The maximum values of pericarp P concentration at 143 DAA were only 15 μmol g⁻¹ pericarp fresh mass in both treatments. A similar accumulation trend between K and P was also evident for content, with no significant increase after reaching 20 μmol per pericarp at 88 DAA in the Standard-K group, and 22 μmol per pericarp at 103 DAA in the Doubled-K fertiliser group (Fig. S4h).

Pericarp Ca concentration remained relatively stable during berry ripening, with values between 3 and 7.5 μmol g⁻¹ pericarp fresh mass in both treatments (Fig. S4i). Pericarp Ca content reached peak values between 6 and 16 μmol per pericarp at 103 DAA in both groups (Fig. S4j).

In Standard-K supply, the pericarp K/Na ratio was greater than 90 throughout ripening. However, the pericarp K/Na ratio was lower than 80 in Doubled-K fertiliser group after 88 DAA (Fig. 1i). ANOVA indicates the K treatment did not significantly alter the pericarp K/Na ratio (P = 0.4898) (Table S2). However, Doubled-K fertiliser decreased the pericarp K/Na ratio after 50 DAA, which appeared to be significantly lower at 88 DAA (P = 0.024) and 143 DAA (P = 0.0136) than the initial value at 50 DAA. By contrast, the K/Na ratio in Standard-K supplied berries did not significantly vary with the ripening phase. On the last sampling date (143 DAA), the Doubled-K supply group tended to have lower K/Na ratio than Standard-K supply group in the pericarp, but this trend was not evident in the rachis or pedicel (Fig. 1j).

Relationships between pericarp parameters in two K supply rates

Shiraz pericarp water percentage and the concentration of five inorganic elements were analysed using PCA to evaluate their distribution patterns during berry ripening as well as the impact of increased K supply on element accumulation (Fig. 2). The combination of principal component (PC) 1 (64.05% variation) and PC2 (18.63% variation) express the majority of data variance (82.68%). Pericarps from the Doubled-K supply group were separated from the Standard-K fertiliser group from 77 DAA onwards, tending to distribute with larger values of PC1 and smaller values of PC2. Ripening berries (DAA: 77–123) were separated from less-ripened (DAA <77) and over-mature berries (143 DAA) by PC2 (Fig. 2a). Over-mature berries (143 DAA) were distinguished from the other berries on PC1 (Fig. 2a). The earlier ripening phases samples were projected into the upper left portion of the scores plot corresponding to the variable loadings projection for higher water percentage and K/Na ratio (Fig. 2b). Besides, samples with higher Mg, P, K and Na concentrations (the ones corresponding to the later ripening phases), were projected in the area defined by positive values on PC1 (Fig. 2b). The distribution patterns agree with the plots for these parameters against with DAA (Figs 1, S3 and S4). The concentrations of Na, K and P were higher in the more mature pericarps than those in the less ripened pericarps. Furthermore, the mature pericarps supplied with doubled K appeared to contain increased Na, K and P than those supplied with standard K. Another set of PCA for pericarp metabolite and element concentrations in earlier and later stages of ripening displayed that less ripened pericarps were characterised by a higher water percentage, L-malic acid, ATP and K/Na ratio (Fig. S5). The variables K/Na ratio and Na concentration were mainly loaded on PC3 (6.84% variation) rather than PC2 (Fig. S5c, d).

Fig. 2.

Principal component analysis for the pericarp water percentage and inorganic element concentrations in the Shiraz K fertiliser trial, including the distribution of samples in the scores plot (a) and the parameters in the loadings plot (b).

Pericarp parameters in ripening Shiraz berries were analysed via correlation coefficient for the Standard-K and Doubled-K supply (Fig. 3), involving four ripening phases, pre-veraison (50 DAA), veraison (63 DAA), harvest maturity (123 DAA) and over maturation (143 DAA). Notably in both treatments, water mass and sugar content were positively correlated, while water percentage and sugar concentration were negatively correlated. Element concentrations and contents were positively correlated with DAA, TSS, pericarp dry mass, as well as sugar concentration and content. By contrast, element concentrations and contents were negatively correlated with water percentage, seed fresh mass, L-malic acid concentration and content, as well as ATP concentration. Evident in only the Standard-K treatment, were a small positive correlation between ATP content and total sugar content, and negative correlations between ATP content and Mg concentration, between berry seed number and Mg concentration, between berry seed number and Na concentration, between pericarp fresh mass and Ca concentration, and between water mass and Ca concentration (Fig. 3a). Positive correlations between pericarp water mass and Na concentration, and between K/Na ratio and seed fresh mass, and a negative correlation between K/Na ratio and Mg concentration were apparent only for the Doubled-K fertiliser group (Fig. 3b). By comparing the correlations of K/Na ratio with other parameters in two treatments (Fig. 3), Doubled-K fertilisation resulted in more significant correlations indicated by the larger dots representing smaller P-values.

Fig. 3.

Comparison of the correlation coefficients for pericarp parameters in Shiraz fertilised with Standard-K (a) or Doubled-K (b) concentration, using samples at 50 DAA (pre-veraison), 63 DAA (veraison), 123 DAA (harvest maturity) and 143 DAA (over maturation). The dots filled with deeper colours represent the more positive (purple) or negative (brown) correlations. The size of the dot represents the absolute value of log₁₀(P), with more significant correlations represented by larger dots. Data with P > 0.05 are not presented. Ethanol concentration and content below the detection limit were removed before the cross-correlation analysis.

Pericarp K concentration was correlated negatively with water percentage (Fig. 4a) and positively with total sugar content (Fig. 4d). However, the relationships between K and other compounds examined in the current study were not always linearly correlated across the ripening phase, for instance, as evident in the clustering of TSS (Fig. 4). Data points with similar TSS, indicated by the colour of circles, clustered in the plots of relationships which were not linearly associated, such as water mass (Fig. 4b), sugar concentration (Fig. 4c), L-malic acid concentration and content (Fig. 4e, f), and ATP content (Fig. 4h), suggesting the relationships between K and these compounds varied in pericarps with different ripeness. The negative correlation between K and ATP concentrations was stronger in pericarps with TSS below 25°Brix. This is indicated by the nonlinear correlation for those data points with TSS higher than 25°Brix (deep purple to dark circles) (Fig. 4g). The relationships between the concentrations of ATP and other components were also dependent on pericarp maturity (Fig. 5), with little influence from the K supply rates.

Fig. 4.

Relationships of K with water (a, b), total sugar (c, d), L-malic acid (e, f), and ATP (g, h) in the Shiraz pericarps treated with Standard-K or Doubled-K fertiliser. Each data point represents one sample. The colour of the data points represents the TSS (°Brix) of the samples. Larger circles or squares indicate a greater pericarp K/Na ratio. Trendlines are fitted using the second polynomial linear regression.

Fig. 5.

Relationships of ATP concentration with water percentage (a), total sugar concentration (b), L-malic acid concentration (c), ethanol concentration (d), Mg concentration (e), P concentration (f), Ca concentration (g), and Na concentration (h) in Shiraz pericarps of vines treated with Standard-K or Doubled-K fertiliser. (d) Ethanol concentrations below the detection limit are not presented. Each data point represents one sample. The colour of data points represents the TSS (°Brix) of the samples. Larger circles or squares indicate higher pericarp K/Na ratios. Trendlines are fitted using the second polynomial linear regression.

Shiraz berry K levels during ripening in own rooted vines or on three different rootstocks

In the rootstock trial, Shiraz berry K concentration and content were affected significantly by DAA (P < 0.0001) and rootstock (P < 0.05) in 2018/2019 and 2019/2020 seasons (Fig. 6) (Table S3). Rootstock also impacted berry fresh mass (P = 0.0007) only in 2018/2019 seasons (Fig. S6) (Table S3). The interactive effect between rootstock and DAA on berry K concentration was significant only in the 2018/2019 season (P < 0.0001), while their interaction on berry K content was significant in both seasons (P < 0.0001). Significant interactions were also observed between rootstock and biological replicate (plant) on berry K concentration, K content and fresh mass in both seasons (Table S3). Berries grown on vines grafted on the 420-A rootstock contained the lowest K concentration and content across two seasons, with top values at approximately 80 μmol g⁻¹ berry fresh mass and 90 μmol per berry. K concentration in berries from vines grown on Ramsey rootstock was relatively low at veraison but was the highest late in ripening in both seasons. On the last sampling date of the two ripening seasons, berry K concentrations were 2.3–4.7% greater for Ramsey rootstock (about 89 μmol g⁻¹ berry fresh mass in the 2018/2019 season; about 88 μmol g⁻¹ berry fresh mass in the 2019/2020 season) compared to own rooted vines (about 85 μmol g⁻¹ berry fresh mass in the 2018/2019 season; about 86 μmol g⁻¹ berry fresh mass in the 2019/2020 season) (Fig. 6a, c). Berry K content was also high for Ramsey rootstock in both seasons. Rootstock Ruggeri-140 carried berries containing the highest K concentrations among the rootstocks and own rooted vines in the earlier ripening phase, until reaching around 75 μmol g⁻¹ berry fresh mass at 81 DAA in the 2018/2019 season, or 101 DAA in the 2019/2020 season. Berry fresh mass tended to be lower for 420-A rootstock relative to the other rootstocks but was comparable to that of own rooted vines (Fig. S6). Ramsey and Rugerri-140 rootstocks produced berries with higher berry mass on all sampling dates in the 2018/2019 season, but the mass of berries grown on these two rootstocks was not significantly higher than those from other rootstocks for most sampling dates in the 2019/2020 season (Fig. S6).

Fig. 6.

Impact of rootstocks on Shiraz berry K concentration (a, c) and content (b, d) in 2018/2019 (a, b) and 2019/2020 (c, d) seasons. Data are presented as mean values with error bars indicating the s.e.m. (n = 10). Significant differences between treatments on the same sampling day are indicated by asterisks (*).

Increased K supply could alter berry composition at harvest

Solutes will be transferred from the roots to the vegetative tissues and eventually into the berries. It is possible that the increased K supply may be absorbed by the vines but could not be further accumulated into berries. Thus, we analysed the five elements in vegetative tissues, petiole, rachis and pedicel (Fig. S2) to better understand the impact of doubling K supply to the vines’ vegetative tissues. It is also necessary to ensure other elements examined in the pericarps did not vary in the vegetative tissues between groups, as this may impact on partitioning to the berry components during ripening. The concentration of elements in vegetative tissues (Fig. S2) indicated the berries received the same amount of most analysed elements from the vegetative tissues, except that K was increased by the Doubled-K treatment. Doubled-K fertiliser increased K and Mg concentration in pericarps simultaneously between 103 and 123 DAA (Fig. 1c, e), suggesting the co-transfer of K and Mg to the berry through the phloem. The increased pericarp Mg concentration in the Doubled-K fertiliser group suggested that increased K availability could promote berry Mg accumulation potentially via translocation from other tissues. Berry hypoxia and greater ethanol production coincided during Shiraz berry ripening (Xiao et al. 2018), and was considered to be important in the physiological transition to berry senescence. In the current study, Mg concentration (Fig. 1e) and ethanol concentration and content (Fig. S4e) increased simultaneously at 123 DAA, possibly associated with the role of Mg²⁺ in cell energisation under O₂ deficiency (Igamberdiev and Kleczkowski 2011). Therefore, doubling K supply may accelerate the import of solutes into berries, subsequently increasing energy requirements. The energisation of the hypoxic berries was potentially assisted by the promoted accumulation of Mg.

The element accumulation pattern in pericarps indicated that extra K supply promoted the accumulation of K and Mg in the mid to late ripening phases (Fig. 1c–e), although Doubled-K nutrition did not impact on most measured ripening parameters, including those relating to berry size and sugar accumulation. The increased K supply also did not intensify berry dehydration, a common occurrence in Shiraz berries late in ripening (Rogiers and Holzapfel 2015) as evidenced by no changes in pericarp water percentage or total sugar concentration. Even though the significant increase in pericarp K concentration and content at 103 DAA and 123 DAA (Fig. 1c, d) was caused by greater K intake by berries, these treatment effects were not apparent at the end of the season. These results are similar to that of a study where K fertiliser was applied only at veraison (56 DAA) and analysed in berries with harvest maturity (Coetzee et al. 2017, 2019). The current study initiated a treatment similar to Coetzee’s study from 50 DAA, followed by repeating the treatment for another six times during berry ripening until over-mature. Although no difference between the two treatments was found in the over-mature pericarps, the K concentration and content on 103 and 123 DAA (Fig. 1c, d), as well as the Mg concentrations on 123 DAA (Fig. 1e), were increased by doubling K supply. The stability of K and other elements in matured berries suggests ionic homeostasis in the later stages of ripening. During the later ripening phases, K partitioning to the berry may rely on remobilised K reserves to make up for the shortfall in newly acquired K, to ensure berry maturation under various external nutrient situations. Doubled-K supply did significantly interact with berry maturity, as evidenced by the interaction between DAA and fertiliser K level (Table S2), to increase pericarp content of L-malic acid at 77 DAA (Fig. 1b). These results are supported by the investigations reporting higher malic acid concentrations in berries from vines fertilised with higher K rates (Marcuzzo et al. 2021). The impact of increased K supply on grape berry solute contents at harvest could be different, depending on the harvest time.

Potential links between pericarp ATP and K levels

Pericarp ATP concentration and content declined considerably during the ripening period in both treatments (Fig. S4c, d), indicating a loss of energisation in the later phases of ripening. The decline is likely the result of increased hypoxia as evidenced by the higher pericarp ethanol and some cell death later in ripening (Fig. S4e, f) and the increased berry ethanol levels during ripening shown in previous studies (Xiao et al. 2018). The negative correlations between ATP concentration and total sugar concentration (Fig. 3) signified that the import of sugar may not be limited by the ATP status decline during most of the berry’s development, supporting the reported high energy efficiency of berry sugar accumulation (Savoi et al. 2021). The strong negative correlations between water percentage and K concentration (Figs 3 and 4a), as well as the positive correlation between sugar content and K content (Fig. 4d), are supported by previous observations of a ternary link between berry sugar, K and water (Coetzee et al. 2017). Additionally, water percentage, water mass, total sugar concentration and content, L-malic acid concentration and content, K concentration and content, and ATP concentration were significantly correlated with each other in both treatment groups, as indicated by the correlation coefficient analysis (Fig. 3). The trendline fitted for ATP concentration against the concentrations of other metabolites and elements were not shifted by increasing K supply (Fig. 5), suggesting that although the K availability may be altered, energy supply and utility within berries was maintained stable and efficient during berry maturation. ATP supply is likely to be involved in regulating solute accumulation and metabolism with secondary effects on water transport in ripening Shiraz berries.

Late in ripening, when TSS exceeded 25°Brix and K concentration was greater than 100 μmol g⁻¹ pericarp fresh mass, the trendline fitted to pericarp ATP and K concentrations switched from a straight line to a curve (Fig. 4g), suggesting a change in the dynamics between ATP content and K accumulation. Severe berry hypoxia late in ripening (Xiao et al. 2018) may limit ATP production, which could be detrimental to membrane energisation. In Arabidopsis thaliana, inhibited K⁺ efflux through GORK channels improved hypoxia tolerance (Wang et al. 2017). Loss of cell vitality in hypoxic berries (Xiao et al. 2018) may be associated with cellular K⁺ transport in later phases of ripening. Further studies assessing the direct impact of K⁺ and ATP on pericarp membrane function during hypoxia are required to clarify these observations and the potential role of ATP in grape berry maturation.

Potential links between pericarp K/Na ratio and ripening progression

Altering the K availability to Shiraz during berry ripening may regulate element redistribution in grapevines. In addition to augmenting K levels between 103 and 123 DAA, Doubling K supply increased the pericarp Na concentration and content across development (P = 0.0183) (Fig. 1g, h), but not in the petiole, rachis or pedicels (Fig. S2). Increased pericarp Na concentration and content also contributed to the reduced pericarp K/Na ratio in the Doubled-K fertiliser group from 88 DAA (Fig. 1i), coinciding with a significant drop in L-malic acid content (Fig. 1b) and almost complete cessation of sugar accumulation (Fig. S4b). In comparison, there was no significant developmental variation in pericarp K/Na ratio in the Standard-K treatment (Fig. 1i). Thus, in vines supplied with Doubled-K fertilisation, the accumulation of Na relative to K may be augmented in berries during the later ripening phases. However, the mechanism for this observation in Shiraz fertilised with extra K is unknown.

Pericarp K content was significantly increased by doubling K supply (Fig. 1d) by the time sugar accumulation was complete (Fig. S4b). However, K concentration and content in this group did not differ from those in the Standard-K fertiliser group on the last sampling date, perhaps indicating a degree of K homeostasis. Besides, there was no significant variation of pericarp L-malic acid between two treatments in the later ripening phase, although there was increased content of L-malic acid in the Doubled-K supply group in the early-ripening phase (Fig. 1b). These observations indicate that K accumulation and L-malic acid exhaustion were more active in the Doubled-K supply group, which might be associated with the reduction of K/Na ratio from the onset of ripening (Fig. 1i). The K/Na ratio was correlated negatively with total sugar concentration and content, and positively with L-malic acid and concentration and content, in both treatments (Fig. 3). Although no obvious sample clustering was apparent in score plots of PC1 versus PC3 (Fig. S5c), an inspection of PC3 indicates a positive loading for Na and a negative loading for K/Na, with the position of K close to the origin on both axes (Fig. S5d). These associations infer that the K/Na ratio was influenced more by the Na concentration than the K concentration in pericarps. The reduced K/Na ratio might impact metabolism, namely since Na⁺ cannot replace K⁺ as a co-enzyme (Maathuis and Amtmann 1999). Thus, increased K supply throughout berry ripening could lead to a reduced pericarp K/Na ratio, correlated with the increased Na concentration in pericarps supplied with higher K (Fig. 1g, h), which indicated that Na⁺ transport may be promoted potentially through K⁺ transporters and channels (Maathuis and Amtmann 1999). The lower pericarp K/Na ratio by increasing K supply may be associated with faster berry maturation and eventually berry senescence, but this needs to be confirmed by further investigations.

Berry K levels were lowest with rootstock 420-A

Shiraz berries from scions on rootstock 420-A contained the lowest K levels in the 2018/2019 season and the 2019/2020 season (Fig. 6), possibly linked with the observation that 420-A reduced K uptake by the scions. This is supported by the study reporting low K levels in vegetative tissues of scions on this rootstock (Hickey et al. 2016). It was also found that Pinot Noir scions grafted on rootstock 420-A had lower K content in petioles sampled at full bloom, compared with those grafted on 101-14 Millardet et De Grasset, 3309 Couderc, 44-53 Malegue, 5 C Teleki, or Harmony (Candolfi-Vasconcelos et al. 1997). The highest berry K concentration was in Ruggeri-140 in the earlier phases of ripening in both seasons, and then in Ramsey after 95 DAA in the 2018/2019 season and 110 DAA in the 2019/2020 season. However, the total K content per berry in Ramsey and Ruggeri-140 was higher than in own-rooted vines, indicating that both Ramsey and Ruggeri 140 rootstocks may somehow promote K accumulation into individual berries. The K concentration in shoots grown on grafted vines was positively correlated with K concentration in xylem sap (Kodur et al. 2010), so K distribution in various tissues may be the result of a combination of xylem and phloem flows. Increased berry K level is predominantly the result of accumulation through the phloem, which is highly active during ripening (Rogiers et al. 2006; Zhang et al. 2006), however xylem flow into Shiraz berries was reduced but remained functional after veraison (Rogiers et al. 2000; Scharwies and Tyerman 2017), thus potentially contributing to berry K accumulation.

Given that DAA significantly altered berry K concentration and content in two seasons (Table S3), berry ripening progression predominated the amount of K accumulated in berries. Therefore, the factors affecting berry ripening progression, such as seasonal variation in weather and other environmental conditions, may affect berry K content. The effect of rootstocks on berry fresh mass varied between growing seasons (Fig. S6, Table S3), indicating that environmental factors interacted with rootstock on berry development. The impact of rootstocks on berry K concentration significantly interacted with the impact of DAA in the 2018/2019 season but not in the 2019/2020 season (Table S3), reflecting the seasonal variation in the rootstocks’ effect on berry K accumulation. Future assessments of the interactions between rootstocks and environmental conditions on berry K accumulation could further characterise the impact of viticultural strategies using rootstocks to manipulate berry ripening.