**3. Results**

#### *3.1. Number of Suckers*

The analysis of deviance showed in all the years that the number of suckers was influenced by the type of treatment, the data collection date, and their interaction (*p* < 0.001, respectively). For each year of the experiment, the least squares means and the standard errors of the number of suckers log transformed are reported in Table 2. Inverse transformed values and lower and upper confidence intervals are reported in Table 3. The number of suckers before the first flaming was similar between treatments in all years. In 2016, all the flaming treatments and hand suckering significantly reduced the number of suckers compared to the initial number of suckers. In 2017 and 2018 only FlamingA (suckers flamed only on 21 April 2017 and 2 May 2018) did not significantly reduce the initial number of suckers. In the control plot, the number of suckers was similar to that before the start of the experiment in all the years (Table 2).


**Table 2.** Least squares means of the number of suckers per plant log transformed and standard errors (SE) as affected by different types of suckering methods and data collection date in the three-year experiment. These data are useful for computing all pairwise comparisons.

FlamingA = plants flamed only on 21 April in 2016 and 2017, 2 May in 2018. The most developed suckers were at the 12–13 BBCH growth stage. FlamingB = plants flamed only on 5 May in 2016 and 2017; 16 May in 2018. The most developed suckers were at the 15–16 BBCH growth stage. FlamingC = plants flamed on 21 April and 5 May in 2016 and 2017; 2 May and 16 May 2018. Hand = plants hand suckered on 21 April 2016 and 2017; 2 May in 2018. Control = no suckered plants.

**Table 3.** Inverse transformed values (from the log scale) and lower and upper confidence intervals (LCI, UCI) of the least squares means of the number of suckers per plant as affected by different type of suckering methods and data collection date in the three-year experiment.


FlamingA = plants flamed only on 21 April in 2016 and 2017, 2 May in 2018. The most developed suckers were at the 12–13 BBCH growth stage. FlamingB = plants flamed only on 5 May in 2016 and 2017; 16 May in 2018. The most developed suckers were at the 15–16 BBCH growth stage. FlamingC = plants flamed on 21 April and 5 May in 2016 and 2017; 2 May and 16 May 2018. Hand = plants hand suckered on 21 April 2016 and 2017; 2 May in 2018. Control = no suckered plants.

In 2016, on 5 May, the number of suckers in FlamingA and FlamingC was significantly lower compared to FlamingB (not ye<sup>t</sup> flamed) and the control, and similar to the hand-suckered plants. On 12 May, all treatments had a lower number of suckers compared to the control. The number of suckers was also lower in FlamingC compared to other treatments, whereas, FlamingA, FlamingB and the hand-suckered plants showed a statistically similar number of suckers. On 9 June, the differences in the number of suckers between treatments were the same as the previous data collection date (four weeks before) (Table 2).

In 2017, on 5 May, the number of suckers in the flamed plants was similar to the nonflamed plants. The hand-suckered plants showed a significantly lower number of suckers compared to the other treatments. On 12 May, the number of suckers in FlamingC, FlamingB and hand-suckered treatments was similar and significantly lower compared to FlamingA and the control (FlamingA and the control were statistically similar). On 9 June, the number of suckers in FlamingA and FlamingB treatments was similar. FlamingA was also similar to the control, whereas FlamingC, FlamingB and the hand-suckering were similar and showed a lower number of suckers compared to the control (Table 2).

In 2018, on 16 May, the number of suckers on the flamed plants was similar to the hand-suckered plants and significantly lower compared to FlamingB and the control. On 23 May, the number of suckers was higher in the control and similar between all other treatments. Flaming and hand-suckering treatments were similar and showed a lower number of suckers compared to the control also on 20 June (Table 2).

In all years, for each flaming treatment and hand suckering, no significant increase in the number of suckers was observed comparing the data collection date after the treatment (two weeks for Flaming A, FlamingC and hand suckering, and one week for FlamingB) and the last data collection date. This thus suggests that in this time period no resprouting of suckers occurred. In all the years, the resprouting of suckers after treatment occurred only during the two weeks between the start of the experiment and the first data collection (i.e., from 21 April to 5 May in 2016 and 2017, and from 2 May and 16 May in 2018) (Table 2).

#### *3.2. Yield Components and Grape Composition*

For each model, all the *p*-values of the analysis of variance are reported in Table 4. The suckering method (treatment) and the treatment:year interaction were not significant, whereas the year was significant for all the dependent variables analysed.

**Table 4.** Analysis of variance for yield, average cluster weight, average berry weight, total soluble solids, pH and tartaric acid, and analysis of deviance (type II Wald chi-square test) for the number of clusters.


Table 5 reports the least squares means and standard errors of the yield, clusters number (log transformed and inverse transformed values with the lower and upper 95% confidence interval), average cluster weight and average berry weight as affected by type of treatment and the year. Within the same year, the yield, number of clusters, and the average weights of the clusters and berries were similar between treatments, suggesting that flame-suckering did not affect yield components.

In the flame-suckered plots, the yield decreased significantly in 2017 compared to 2016, and in 2018 compared to 2017. In the hand-suckered and control plots, the yield was similar in 2016 compared to 2017, whereas in 2018 was lower compared to 2016 and 2017. The average berry weight in 2016 was similar to 2017, whereas it was higher in 2018 compared to 2016 and 2017 for all the treatments. Except for the hand-suckered plants, where the number of clusters per plant was similar in 2016 and 2017, for the other treatments, the number of clusters increased from 2016 to 2017. The number of clusters decreased significantly from 2017 to 2018 for all the treatments. The average cluster weight was significantly lower in 2017 and 2018 compared to 2016 and was similar in 2017 and 2018 for all the treatments.


**Table 5.** Least squares means and standard errors of the yield, clusters number (log transformed) average cluster weight and average berry weight as affected by the type of treatment and the year. Inverse transformed values and lower and upper 95% confidence intervals (LCI, UCI) of clusters number log transformed are reported.

FlamingA = plants flamed only on 21 April in 2016 and 2017, 2 May in 2018. The most developed suckers were at the 12–13 BBCH growth stage. FlamingB = plants flamed only on 5 May in 2016 and 2017; 16 May in 2018. The most developed suckers were at the 15–16 BBCH growth stage. FlamingC = plants flamed on 21 April and 5 May in 2016 and 2017; 2 May and 16 May 2018. Hand = plants hand suckered on 21 April 2016 and 2017; 2 May in 2018. Control = no suckered plants.

Least-squares means and standard errors of the total soluble solids, pH and tartaric acid as affected by the type of treatment and the year are reported in Table 6. The values of total soluble solids, pH and tartaric acid were statistically similar between treatments in all the years. In 2016, the total soluble solids and pH values were significantly lower compared to values estimated in 2017 and 2018 for all the treatments. The amount of tartaric acid was significantly higher in 2016 compared to 2017 and 2018 for all the treatments. The total soluble solids, pH and tartaric acid content were similar in 2017 and 2018 for all the treatments. As total soluble solids increase in the berries, the juice pH rises and the tartaric acid declines.


**Table 6.** Least-squares means and standard errors of the total soluble solids, pH, and tartaric acid as affected by the type of treatment and the year.

FlamingA = plants flamed only on 21 April in 2016 and 2017, 2 May in 2018. The most developed suckers were at the 12–13 BBCH growth stage. FlamingB = plants flamed only on 5 May in 2016 and 2017; 16 May in 2018. The most developed suckers were at the 15–16 BBCH growth stage. FlamingC = plants flamed on 21 April and 5 May in 2016 and 2017; 2 May and 16 May 2018. Hand = plants hand suckered on 21 April 2016 and 2017; 2 May in 2018. Control = no suckered plants.

#### *3.3. Flaming Machine Performance and Costs*

Flaming machine performance and costs are reported in Table 7. It should be pointed out that a large amount of the variable costs for conducting flaming was due to the high cost of the LPG in Italy (2.25 € kg−1), and that flaming may be less expensive in countries where the LPG (or propane) would costs less (e.g., in the USA the propane cost is equivalent to 0.48 € kg−1).



\* Time to conduct the operation in a hypothetical area of 10,000 m<sup>2</sup> (30.00 m wide and 333.33 m long).
