**1. Introduction**

Suckers are nonbearing shoots that grow in the spring from latent buds on grapevine (*Vitis vinifera* L.) trunks [1]. Sucker growth can lead to excess vegetation, increase the possibility of attack from pathogens and alter the fruit/shoot ratio [2]. Moreover, suckers can cause problems during vineyard managemen<sup>t</sup> operations, such as soil tillage, weed removal, mechanical harvest, and pest and disease control [3]. To overcome these problems suckers are removed during grapevine cultivation and this process is known as suckering. The right time for suckering is when they are not ye<sup>t</sup> lignified. Waiting longer causes the suckers to become lignified, harden, which are then more difficult to remove. Suckering in spring also prevents the development of resprouting basal buds [4].

Traditionally, suckering was done by hand, however this is costly and time consuming because it requires constant bending down, getting up and making repetitive motions [5]. Hand suckering requires an operating time ranging from a minimum of 20 h ha−<sup>1</sup> to a maximum of 60-70 h ha−1, depending on the operating conditions [6]. The mechanical removal of suckers by scourges is widely employed, however this is generally stressful on young plants, which can be damaged by rotating scourges [2]. Chemical suckering with traditional herbicides or synthetic growth regulators is also widely used [2,7], however the use of synthetic chemicals is forbidden for organic wine grape growers.

Flaming could be a viable nonchemical alternative to remove the not ye<sup>t</sup> lignified spring suckers. The high temperature of the flame denaturises the plant proteins of green tissues, without burning, and

thus desiccates them [8]. Flame-suckering could be useful for organic viticulture, which has received increased interest by grapevine growers in the recent decades.

Flaming is currently used to control weeds in heat-tolerant herbaceous and horticultural crops [9–12], however, to the best of our knowledge, there has been no research using flaming to remove suckers from grapevines. This research tests the effects of flaming to remove the suckers. Grape yield components and grape composition were also recorded.

#### **2. Materials and Methods**

#### *2.1. Experimental Set Up*

A three-year experiment (2016, 2017 and 2018) was conducted on a 10-year-old Sangiovese vine (clone BF-30) grafted on 775 Paulsen rootstock. The farm (Tenuta Ceppaiano, Castellani Spa) was located in Tuscany, Italy (43◦3551.6 N 10◦3213.8 E). The vineyard training system was spurred cordon. The cordons were 80 cm height. The distance between each vine on the row was 80 cm, and between the rows was 2.10 m, for a density of 5952 plants per ha. The soil was loam (40% sand, 34% silt, 26% clay, 1% organic matter, pH = 7). Figure 1 reports the monthly-cumulated rainfall and monthly average temperatures recorded during the three-year experiment. Fertilization consisted of the application of an organic-mineral fertilizer in January 2016 and 2017 (10N-5P-14K and 8N-16P-24K, respectively), and calcium nitrate (15.5N–0P–0K) in January 2018. Sixteen, eight and eleven chemical treatments, against *Plasmopara viticola* (Berk. & M.A. Curtis) and *Uncinula necator* (Schwein.) Burrill, were applied from April to August in 2016, 2017 and 2018, respectively. One chemical treatment against *Lobesia botrana* (Schiff. et Den.) was applied in June in 2016 and 2018. The vineyard was not irrigated.

**Figure 1.** Monthly cumulated rainfall and monthly average temperatures (January 2016–December 2018) recorded by the meteorological station in Siberia, Crespina-Lorenzana (Pisa, Italy) (43◦3531.2 N 10◦3238.4 E) [13].

The flaming machine used for suckering was the PFV-600 model (Officine Mingozzi, Ferrara, Italy) [14] (Figure 2). A mobile horizontal frame supports the burners, which are placed in two rows in a staggered position. The inclination of the burner rows can be adjusted based on the height of the suckers. The burners were cylindrical with air-intake and operated in gaseous phase (Figure 2) [15]. The flaming machine was coupled with a SAME Frutteto 100 (Same, Treviglio, Bergamo, Italy) tractor.

**Figure 2.** The flaming machine PFV-600 (Officine Mingozzi, Ferrara, Italy) coupled with a SAME Frutteto 100 (SAME, Treviglio, Italy) tractor.

The flaming machine was supplied with liquefied petroleum gas (LPG). The LPG consumption at the pressure of 0.2 MPa was 17.64 kg h−1. The working width was 1.05 m (half of the 2.10 m inter-row space) and the forward speed was 3 km h−1. The machine distributed 55.90 kg ha−<sup>1</sup> of LPG. The LPG dose actually applied to the suckers (within the intra-row space of 0.30 m) was 195.65 kg ha−1. This LPG dose was chosen because it was deemed effective to devitalize suckers on the basis of previous experiments where flaming was used to devitalize weeds and cover crops [10,11,16–19].

Each year, the first flaming was applied in the spring when the vine plants showed the most developed suckers at the 12–13 BBCH growth stage (two–three unfolded leaves) and the main productive shoots at the 15–16 BBCH growth stage (five–six unfolded leaves) [20] (21 April in 2016 and 2017, 2 May in 2018) (Figures 3 and 4) (Supplementary Materials). The second flaming was applied when the nonflamed plants showed the most developed suckers at the 15–16 BBCH growth stage and the main shoots at the 18-19 BBCH growth stage (eight–nine unfolded leaves) (5 May in 2016 and 2017, 16 May in 2018). Hand suckering was conducted on the same date as the first flaming. The control was not suckered in the spring, but lignified suckers were manually removed during the winter pruning. The suckers that remained on the plants after the flaming and hand suckering, or that had resprouted, were also removed during the winter pruning.

**Figure 3.** Flame suckering applied on 21 April 2017 at the 13 BBCH sucker growth stage.

**Figure 4.** Flame suckering applied on 21 April 2016 at the 12–13 BBCH sucker growth stage.

The experimental design was a randomized complete block design. Five adjacent vineyard rows were selected, and each row was divided into five 16-m-long blocks (one for each treatment). Treatments were: (1) flaming applied once only when the most developed suckers were at the 12–13 BBCH growth stage (treatment "FlamingA"), (2) flaming applied once only when the most developed suckers were at the 15–16 BBCH growth stage (treatment "FlamingB"), (3) flaming applied twice, the first time when the most developed suckers were at the 12–13 BBCH growth stage, and the second time at the same date as FlamingB (treatment "FlamingC"), (4) hand suckering when the most developed suckers were at the 12–13 BBCH growth stage (treatment "Hand"), and (5) nonsuckered plants (treatment "Control") (Table 1).


**Table 1.** Date of suckering treatments, and the growth stage of the most developed suckers at the time of suckering.

## *2.2. Data Collection*

Each year, data were always collected in relation to the same five vine plants at the centre of each block for a total of 25 replicates for each treatment. The persistence of suckers after treatments was evaluated by counting the number of suckers at four different times: (1) immediately before the first flaming, (2) two weeks after FlamingA, (3) three weeks after FlamingA, and one week after FlamingB, and (4) seven weeks after FlamingA, and five weeks after FlamingB.

In September, at the harvest, all the clusters of each replicate were counted and weighed together in order to evaluate the yield. The average cluster weight (g cluster−1) was calculated by dividing the yield by the number of clusters. The average berry weight (g berry−1) was calculated by averaging the weight of 50 berries randomly picked from the clusters of each replicate.

Immediately after harvest, the berries were placed in hermetically sealed plastic bags and stored in a cooler at 4 ◦C to preserve their characteristics. The berries were then crushed and the juice filtered through cheesecloth to determine total soluble solids, pH and tartaric acid following standard methods (European Commission Regulation (EC) No. 2676/90). Total soluble solids (Brix) were determined at 20 ◦C using an ATC digital refractometer (Tekcoplus, Hong Kong, China); pH was measured using a Hanna H18519N electronic pH-meter (Hanna Instruments, Padova, Italy); and tartaric acid was determined by acid-base titration using sodium hydroxide (0.1 N) to an endpoint pH of 8, with values expressed as tartaric acid (g <sup>L</sup>−1).

Flaming machine performance parameters and costs were calculated. The field efficiency (i.e., the ratio of the theoretical field time and the total time spent in the field) was computed by referring to a hypothetical area of 10,000 m<sup>2</sup> (30.00 m wide and 333.33 m long). The theoretical field time is the time the machine is effectively operating at an optimum forward speed and performing over its full width of action. The total time for conducting the operation was calculated by summing the machine adjustment time (including plugging and unplugging), the theoretical field time, the turning time, and the time to refuel the tractor and/or replace empty LPG tanks. However, the travelling time back and forth the field was not included. The total cost per use was calculated by summing the fixed and variable costs for the flaming machine coupled with a SAME Frutteto 100, following a standard methodology for cost determination [21]. The rate of depreciation was determined considering a purchase price of €46,445 for the SAME Frutteto 100, and € 12,139 for the flaming machine. The economic lifetime considered was 12 years for the tractor, and 10 years for the flaming machine. The repairing and maintenance factor was 80% for the tractor, and 75% for the flaming machine. The labour costs for the tractor driver was 15 € <sup>h</sup>−1, and the LPG cost was 2.25 € kg−1.

## *2.3. Statistical Analysis*

Data normality was assessed using the Shapiro–Wilk test. Other tests consisted of the Student t-test to verify that the mean error was not significantly different to zero, the Breusch–Pagan test for homoscedasticity and the Durbin–Watson test for autocorrelation.

Data on the number of suckers were modelled in a generalized linear mixed model using the extension package lmerTest (Tests in Linear Mixed Effects Models) [22] of R software [23]. The log transformation was assessed. The treatments and data collection dates were the fixed factors. The random factors (replicates and data collection dates) were assessed as longitudinal data (repeated measures) to obtain a correlated random effect for intercept and slope. Data were analysed separately each year. The analysis of deviance was run. The package emmeans (Estimated Marginal Means, aka Least-Squares Means) [24] was used to compute the least squares means, standard errors, inverse transformed values, and confidence intervals.

Yield components and grape composition data were modelled in a linear mixed model using the extension package lmerTest [22] of R software [23]. Treatments and years were the fixed factor, and replicates and years were the random factors. The analysis of variance was run. The package lmerTest [22] was used to compute the least squares means and standard errors.

Pairwise comparisons between estimated least squares means were computed by estimating the 95% confidence interval of the difference between the least squares means (Equation (1)):

$$\text{CI (difference)} = (\mathbf{x}\_1 - \mathbf{x}\_2) \pm 1.96 \sqrt{\left(\text{SE}\_{\mathbf{x}\_1}\right)^2 + \left(\text{SE}\_{\mathbf{x}\_2}\right)^2} \tag{1}$$

where (*x*1) is the mean of the first value, (*x*2) is the mean of the second value, (SE*x*1) is the standard error of (*x*1), and (SE*x*2) is the standard error of (*x*2) [25].

If the resulting 95% confidence interval (CI) of the difference between values did not cross the zero value, the null hypothesis that the compared values were not different was rejected.
