Measurements were carried out in a 1990 planting of Semillon (clone 32) and Shiraz (clone BVRC 12) on their own roots at the University of Adelaide (Waite Campus, South Australia, 34°58'3.47"S; 138°38'0.43"E) over a period of four growing seasons for Semillon, from 2014 to 2018 (season one (S1), 2014–15; season two (S2), 2015–16; season three (S3), 2016–17 and season four (S4), 2017–18), and two seasons for Shiraz, from 2016 to 2018 (S3 and S4). Both varieties were trained to a bilateral spur pruned cordon (two buds per spur) with vertically positioned shoots. Semillon vines were planted at a 3 × 1.8 m spacing (averaging 17 spurs/vine) and Shiraz to a 3 × 2.7 m spacing (averaging 22 spurs/vine). The blocks were irrigated with an average of 0.5 ML/ha.

A control (C), where no canopy interventions were carried out and a bunch thinning (BT) application performed at the onset of veraison, EL 33 (Coombe, 1995) (~10 °Brix) were compared. A fully randomised design was set up across three rows of Semillon, where each treatment was repeated once along each row in blocks of nine consecutive vines. Measurements were carried out on three vines per treatment per row. A different design was used for Shiraz, where own-rooted vines were selected on nine rows of a pre-existing rootstock trial. In each row, one vine per treatment was selected (n = 9). For the BT treatment, all bunches on the vines included in the trial were counted, on a per m basis, and exactly 50 % of them were removed in both varieties; distal, small and tangled ones were preferentially removed to emulate industry practice.

The Kent Town weather station (station number 23090) of the Australian Bureau of Meteorology (

http://www.bom.gov.au/

) collected temperature and rainfall data. Hall and Jones (2010) have previously described the climate of the Adelaide Plains region, where this vineyard is located, as hot (minimum GDD = 2072, maximum GDD = 2209).

The mesoclimate at the vineyard site during the four growing seasons (October to April) has been previously described in De Bei et al. (2019) and De Bei et al. (2020). The growing degree days (GDD) (Gladstones, 2011) were 1811, 1899, 1770 and 1886 in S1, S2, S3 and S4, respectively, compared to a long-term average (LTA) of 1801. The LTA growing season rainfall was 198.4; during the four seasons of the trial, it varied from 137.8 mm (S1) to 354.2 mm (S3).

Figure 1 depicts the minimum and maximum daily temperature and daily rainfall of the four growing seasons (October to April) of the trial. Season three was the wettest, with a peak of 61.2 mm of rain on the 28th Dec 2016; it was also the coldest of the four seasons.

Table

Wines were made from 20 kg of fruit from each treatment replicate. The vineyard replicates were maintained to produce three wines per treatment. The winemaking procedure is described in detail in De Bei et al. (2019). The finished wines were bottled into 375 mL crown sealed glass bottles and stored in a temperature-controlled environment (18 °C) until further analysis.

On the finished wines, pH and TA were measured with an autotitrator (Mettler Toledo auto titrator, Greifensee, Switzerland), and alcohol (v/v) content was measured using an Alcolyzer Wine ME (Anton Paar, Graz, Austria). The method proposed by Iland et al. (2004) was applied to the measurement of total phenolics, while total tannins concentration was measured according to Mercurio et al. (2007) via the methylcellulose precipitate (MCP) assay.

Sensory assessment of berries (BSA) and wines were carried out as previously described in De Bei et al. (2019) and De Bei et al. (2020) and according to the method proposed by Olarte Mantilla et al. (2013). For the BSA, from the harvested fruit, a total of 300 berries per variety and treatment replicate were kept at 4 °C until assessment, which was carried out within one week. Attributes of the pulp (P), skin (Sk) and seeds (S) were assessed. A number of 14 attributes were assessed for Semillon as follows: P-juiciness, P-acidity, P-citrus flavour, P-tropical flavour, P-grassy flavour, P-flavour intensity, Sk-acidity, Sk-bitterness, Sk-astringency, Sk-grape flavour, Sk-grassy flavour and S-colour, S-flavour and S-astringency. The attributes assessed for Shiraz were: P-sweetness, P-acidity, P-dark fruit, P-red fruit, P-dried fruit, P-green/grassy, P-juiciness, Sk-acidity, Sk-bitterness, Sk-astringency, Sk-grassy flavour, Sk-dark fruit flavour, Sk-red fruit flavour, S-flavour, S-bitterness and S-astringency.

Assessors (10–12 depending on the season) were trained for two 2-hour sessions before the three formal assessment sessions required to complete the BSA.

Similarly, wine sensory assessment was carried out using descriptive analysis (DA) as described in De Bei et al. (2019). Panellists were trained using ranking exercises for attributes such as acidity, astringency and bitterness. Assessors were also asked to identify unknown aroma standards as a training exercise. In total, 22 attributes were assessed in each of the formal sessions for Semillon and 20 for Shiraz. Attributes were divided into categories as follows: aroma (Semillon: confectionery, citrus, tropical, grassy, intensity; Shiraz: dark fruit, red fruit, jammy, pepper, green/grassy, aroma intensity), taste (bitterness, acidity and sweetness), flavour (Semillon: bitter, acid, citrus, floral, stone fruit, confectionery, tropical, grassy; Shiraz: dark fruit, red fruit, jammy, pepper, green/grassy and spice), mouthfeel (body, astringency, alcohol) and aftertaste (fruit length, alcohol length, bitter length, likeability).

Both BSA and wine DA were approved by the University of Adelaide ethics committee (H2017-054) and took place in a sensory facility of the University of Adelaide at the Waite Campus.

Treatment comparisons, seasonal effects and the interaction of the two (treatment x season) were analysed via two-way ANOVA; means were then separated using the Tukey’s test. Sensory results were assessed using the XLSTAT® product and sensory panel performance analysis tool. XLSTAT® (Version 2015.4.01.20116 Addinsoft SARL, Paris, France) was used for all statistical analyses.

Means were separated at a significance level of p < 0.05 for all data but the berry and wine sensory for which a significance of p < 0.1 was considered (Olarte Mantilla et al., 2013).

Bunch thinning, as intended, reduced the number of bunches per m compared to C in both varieties (Table 1). This had a strong effect on yield, which was lower in BT in all seasons and varieties (Table 1 and supplementary tables S1 and S2). For Semillon, the greatest difference was observed in S1, where BT yielded 47 % less than C, while in S2, the difference was 29 %, the lowest in the whole trial (Table S1). For Shiraz, the yield was 50 % and 40 % lower in BT in S3 and S4, respectively (Table S2). Bunch and berry weight were not affected by BT apart from Semillon in S3 when they were both heavier in BT (Tables S1 and S2). S3 is the main driver of the general difference observed in bunch weight in Semillon (Table 1).

Semillon showed a smaller leaf area in BT; however, no differences were then observed in the pruning weight measure. For Shiraz, no differences were observed both in leaf area and pruning weight (Table 2). The LA/Y of Semillon, for all seasons and treatments, varied between 1.2 and 3.4. Given the much lower yield observed in BT in all seasons, it is not surprising that the LA/Y was higher in BT. On the other hand, the Y/P showed a wider variation (ranging from 4.7 to 15) and no differences between treatments in Semillon. For Shiraz, the LA/Y was higher in BT while the Y/P was greater in C (Table 1). Strong seasonal influences emerged for most of the yield component results in Semillon, while for Shiraz, only the bunch weight and leaf area were influenced by the season.

Table

titre du tableau
		Control	BT	Season	Treatment × Season
Semillon	Bunches (#/m)	36.7 a	21.3 b	ns	ns
	Yield (kg/m)	6.4 a	4.1 b	0.003	ns
	Bunch weight (g)	177.5 b	196.7 a	0.0001	ns
	Berry weight (g)	1.7	1.6	0.001	ns
	Leaf area (m2/m)	9.1 a	8.4 b	0.0001	ns
	Pruning weight (kg/m)	0.77	0.69	0.001	ns
	LA/Y (m2/kg)	1.5 b	2.3 a	0.0001	ns
	Y/P	10.6	7.5	ns	ns
Shiraz	Bunches (#/m)	28.1 a	16.1 b	ns	ns
	Yield (kg/m)	4.4 a	2.5 b	ns	ns
	Bunch weight (g)	158.1	156.7	0.009	ns
	Berry weight (g)	1.6	1.5	0.03	ns
	Leaf area (m2/m)	7.7	7.7	0.010	ns
	Pruning weight (kg/m)	1.5	1.3	ns	ns
	LA/Y (m2/kg)	2.1 b	3.3 a	ns	ns
	Y/P	3.2 a	1.9 b	ns	ns

C = control, BT = Bunch Thinning, LA = leaf area, Y = yield, P = pruning weight. Treatment and season effects and their interactions were analysed using a two-way ANOVA and Tukey’s test. Means followed by different letters are different at p < 0.05. ns = not significant.

For the variety Semillon, in the first two seasons of the trial, differences were measured at harvest when, in S1, TSS was higher in BT and, in S2, berry weight was greater in BT. In S3, TSS was greater in BT in each sampling date, which brought forward the C’s harvest by six days to ensure similar final TSS in both treatments. From DOY 55 to DOY 61, the sugar accumulation in C was 0.62 Brix/day, if this rate was maintained, C should have been harvested at DOY 59 to achieve the same TSS as BT. In S4, no differences were measured in either the sugar accumulation and berry growth in the two treatments. Similarly to what was observed in the previous season, in S4, BT was harvested four days earlier than C.

Table

Treatment effects were analysed using a one-way ANOVA and the means separated with Tukey’s test. Means followed by different letters are different at p < 0.05. ns = not significant.

Differences in the sensory profile of the Semillon berries were found in all four seasons (Figure 1). The tropical flavour of the pulp was higher in S3 and lower in S1 and S4 for BT. The grassy flavour of the pulp was different in all seasons, and, apart from S2, berries were more green/grassy in BT. Similarly, in S1 and S3, the skin of C was also grassier. BT had the juiciest pulp in S1 and S3 but not in S4. The acidity of BT pulp was higher in S3 and S4, while the skin was less acidic in S3. BT skin was less bitter than C in three out of four seasons; similarly, the skins were less astringent in S2 and S3.

Table

a: season one (S1), 2014–15; b: season two (S2), 2015–16; c: season three (S3), 2016–17; d: season four (S4), 2017–18. P = Pulp, Sk = Skin, S = Seeds, Fl = Flavour.

Bunch thinning carried out at veraison by removing as much as 50 % of the bunches had little effect on the performance of the Semillon variety in a hot Australian climate, and indeed, wine produced from the untreated control were preferred. The effect on Shiraz was more notable, especially in the final wine, to confirm that the practice should be considered in managing red varieties rather than white. To support the decision on whether to bunch thin and justify its cost, a significant increase in fruit quality has to be achieved. Keller et al. (2005) discussed how the practice is often dictated by wineries, based on the belief that higher yield would result in poor quality wines even though evidence of this relationship is limited and most often based on data from cool climate viticulture. To reiterate this, yields in European countries are regulated by law so that often, techniques such as bunch thinning are non-negotiable to remain within the allowed yield per hectare rather than being driven by a quality target. In the early 1990s, Jackson and Lombard (1993) advocated for the need to investigate the optimum yield for each cultivar in each region and training system to produce quality wine and remain profitable. This study contributes to that goal and has demonstrated that bunch thinning of Semillon in a hot climate did not alter measurable chemistry parameters despite considerable (especially from an economic point of view) differences in yield. Moreover, despite the differences in sensory attributes of the grapes and wines produced from control and bunch thinned vines, it was difficult to extract a clear pattern; often, the more productive control vines resulted in wines that were judged more favourably by a panel of assessors in their ability to express better the expected varietal characters of Semillon grapes. For Shiraz, no gains in berry chemistry were observed; however, a better performance of the BT wines during the sensory analysis was observed.

Bunch thinning, by reducing yield, is also likely to promote a faster ripening of the grapes (observed in some seasons of this study both in Semillon and Shiraz). With recent vintages showing a trend of early harvests and significant compression of the ripening window (Jones et al., 2005; Palliotti et al., 2014), bunch thinning is likely to exacerbate this issue.

Further research on other high yielding varieties (particularly reds) and in commercial vineyards, where bunch thinning is unlikely to remove 50 % of the fruit, is required to understand better the effect of this management technique in hot climates.

This research was supported by funding from Wine Australia. Wine Australia invests in and manages research, development and extension on behalf of Australia’s grape growers and winemakers and the Australian Government. Thanks to all of the Viticulture laboratory staff and interns at the University of Adelaide, in particular Ms Annette James, who assisted in data collection. The authors also acknowledge all the sensory assessors who participated in the study. Special thank you to the Coombe vineyard’s staff, particularly Mr Phil Earl and Mr Ben Pike for their support with the trial.