Manipulating the Postharvest Period and Its Impact on Vine Productivity of Semillon Grapevines

Vineyards and treatments.

Trial sites were established in four Vitis vinifera L. cv. Semillon (on their own roots) vineyards that differed in cropping level and postharvest growing conditions. Two of the sites were chosen in high-yielding, furrow-irrigated vineyards in the Riverina region (34°S, 146°E), and two in lower-yielding, drip-irrigated vineyards in the neighboring Hilltops region (34°S, 148°E) of New South Wales, Australia. The Riverina has been classified as a very hot grapegrowing region with a mean January temperature of 24.2°C (Dry and Smart 1988) and mean annual rainfall of 406 mm. Annual rainfall in the Hilltops region is 653 mm, and the mean January temperature is 22.9°C, placing it in the hot-climate category (Dry and Smart 1988). All vineyards were trained to a single-wire cordon, while planting density varied among vineyards (Table 1⇓). The Hilltops vineyards (H1, H2) were at a higher elevation than the more recently planted Riverina vineyards (R1, R2). The phenological dates of the four sites indicate that the season is slightly longer in Riverina and harvest date was also earlier in the first season, resulting in a longer postharvest period. This is even more apparent in the postharvest heat accumulation (growing degree days [GDD] base 10°C) of that season; the Riverina vineyards accumulated considerably more heat after harvest (Table 1⇓). The GDD were calculated from budburst to the end of the growing season (leaf fall). Rain and irrigation amounts were much lower in the Hilltops vineyards, especially during the postharvest period. Only vineyard H1 received some irrigation water in the 2001/2002 season after harvest, while vineyard H2 received no additional water during postharvest. In the last season of the study, H2 was not irrigated, and there was no commercial harvest. The soil was sandier in H2, which could have reduced the availability of water, thereby increasing the impact of the drought conditions in the 2002/2003 season. Under commercial operation all vineyards were machine-harvested. Vineyards R2 and H1 were hand-pruned, and vineyards R1 and H2 were machine-pruned, but vineyard R1 received a follow-up hand-pruning.

At each site, six treatments were imposed over two consecutive seasons to change either the length of or accumulation level during the postharvest period. Relative to control vines, four of these treatments were designed to reduce the amount of carbohydrate reserves accumulated after harvest, and one treatment was designed to increase reserve accumulation. The treatments were as follows: untreated control (hand-harvested at commercial maturity); crop removal (harvested several weeks before harvest in other treatments); leaf removal (all leaves removed at harvest); leaf damage (a simulated 50% loss of leaf area by machine-harvesting); juice spray (a simulated effect of juice splashing on leaves by machine-harvesting); and leaf damage plus juice spray (combined simulated leaf loss and juice splashing). Under the crop-removal treatment, the grapes were removed during the phase of rapid berry sugar accumulation, when fruit sink strength is at its maximum (Candolfi-Vasconcelos et al. 1994). The crop-removal treatment was applied about 5 weeks (2000/2001) and 7 weeks (2001/2002) before the commercial harvest on average for each region. Total soluble solids (TSS) at the time of crop removal was recorded only in 2001/2002 and was 7.4 Brix in the Riverina vineyards and 12.2 Brix in the Hilltops vineyards. In the second season the leaf-removal treatment required repeated defoliation after harvest because of regrowth of shoots. Leaf damage was inflicted at harvest in 2001 by a commercial harvester followed by a lawn trimmer to ensure similar damage in all vineyards. At harvest 2002, the leaf damage was done manually by hitting the canopy with two 1-m-long wooden sticks to simulate the effect of the mechanical beaters on the machine harvester. Approximately 50% of the leaves were removed based on visual assessment, with this level of damage confirmed by subsequent aerial imaging as described by Hall et al. (2004). For the juice-spray treatments, grape juice was filtered through a cloth and applied with a 10-L garden sprayer. Between 5 and 10 L was sprayed on the leaves of each five-vine replicate, covering the whole canopy until runoff.

All treatments were replicated four times on five consecutive vines in a fully randomized block design and implemented over two growing seasons (2000/2001/2001/2002). The data and sample collection started at harvest in 2001 and concluded shortly after pruning in 2003, but only data from the last two full growing seasons (2001/2002/2002/2003) are reported here, since the treatments would have only altered vine productivity and grape composition in those seasons.

Vine and berry measurements.

Yields were determined by hand-harvesting the three middle vines in each treatment plot just before the commercial harvest (except H2 in 2003). In the 2001/2002 season the Riverina vineyards were harvested several weeks earlier than the Hilltops vineyards, while in the following season (2002/2003) all vineyards were harvested in February (Table 1⇑). The clusters were counted at harvest in the 2001/2002 season and about four weeks after budbreak in the 2002/2003 season. A 100-berry sample was taken at harvest from both sides of the canopy to determine berry weight and TSS with a digital bench refractometer (PR-101, Atago, Tokyo, Japan). Juice pH and titratable acidity (TA) were determined using an automatic titrator (TitraLab 80, Radiometer Copenhagen, Lyon, France). Berries per cluster, cluster weight, and sugar per berry (TSS x total berry weight) were calculated from these data. The canes were counted at pruning, and the average cane weight was calculated using the weight of prunings from each vine. All vines included in each trial (treatment vines) were pruned to equal bud numbers, but otherwise the vines were hand-pruned to match the surrounding commercial vineyard. Vineyard R1 was pruned to ~60 buds per vine; R2 and H1 to ~50 buds per vine; while H2 was pruned to ~110 buds per vine. Trunk circumference was measured in winter with a measuring tape at two positions (30 cm from the vine base and 30 cm below the wire) from each sampling vine in a replicate.

Data analysis.

Statistical analysis was conducted using GenStat release 6.1 (VSN, Herts, UK). An analysis of variance was applied separately for each site to detect differences among the treatments. Differences between means were detected using a least significant difference (LSD) test (p = 0.05).

Yield and pruning weight.

Vine productivity was affected by crop removal, and leaf removal at all sites except H2, but the magnitude of the effects varied between regions, perhaps because of the differences in productivity levels (Table 2⇓). In the higher yielding Riverina vineyards the impact on yield and pruning weight was great, while in the Hilltops vineyards the changes were small or not detectable. The yields in vineyard R1 were affected in both years; the yield reduction because of leaf removal was ~20% in 2002 and ~50% in 2003. In contrast, yields were increased by 20% when the crop was removed early, while leaf damage had no effect on yield. Yields in vineyard R2 were reduced by leaf removal or increased by crop removal by ~30% at harvest 2003, and the other treatments had minor effects as in vineyard R1. There was a similar response pattern in the same year in vineyard H1, where these were decreased or increased ~40% by leaf and crop removal, respectively. There was no response to treatments in vineyard H2, probably because of the drought conditions since the vineyard received no irrigation during the 2002/2003 season or in the previous postharvest period. In addition, the sandier soil in vineyard H2 would have retained less soil water from winter rains that would be available to vines during the growing season. Surprisingly, spraying juice on the leaves did not reduce vine productivity, but rather increased yield in 2003 in one of the Riverina vineyards because of larger clusters. Despite being sprayed to runoff with grape juice, the spray did not induce the leaf burning sometimes seen after machine-harvesting in Semillon.

The reduction in yield in the seasons following complete defoliation at harvest and the increasing effect over multiple seasons have been reported previously (Scholefield et al. 1978), suggesting an impact on all key stages of reproductive development. As in our study, the authors found only a minor impact of partial defoliation on yields. The cumulative effect of defoliation during berry development applied in previous seasons on yield was also demonstrated by Candolfi-Vasconcelos and Koblet (1990), who found concomitant reduction in starch reserves. According to Loescher et al. (1990), late-season defoliation always results in diminished carbohydrate reserves and can also result in nitrogen deficiency in the following season. Complete defoliation not only would have prohibited nutrient resorption from the leaves and reallocation to storage (that is, it interfered with the normal senescence process in autumn), but also would have greatly reduced late-season nutrient uptake from the soil because of the elimination of transpiration. These vines would therefore also have a reduced nutrient status (especially for nitrogen), as indicated by lower nutrient levels in the spurs at leaf fall 2002 (Holzapfel and Smith 2005). Both vegetative growth and fruiting of grapevines may be more dependent on nitrogen reserves than on carbohydrate reserves (Cheng and Xia 2004). Increasing reduction in yield over multiple seasons could be due to a progressive decline in vine reserve levels, which may impact all stages of reproductive development. The minor impact of the damage of the simulated mechanical harvester could have been due to a compensatory increase in photosynthetic activity of the remaining functional leaves, as shown for partial defoliation (Hunter and Visser 1988), or to improved light interception by the remaining leaves. Alternatively, the remaining 50% of the leaves may have been sufficient to maximize yield potential for the following season.

In addition to the higher crop levels, the Riverina vineyards had much higher pruning weights than the Hilltops vineyards and were more affected by the treatments in both years (Table 2⇑). The Riverina vineyards responded differently in the first year (2002); pruning weights were lowered by leaf removal in vineyard R1 and elevated by crop removal in vineyard R2. In the following season pruning weights were reduced by leaf removal and elevated by crop removal similarly in both vineyards. Crop removal also increased pruning weights in vineyard H1 but had no effect in vineyard H2, where drought stress led to low pruning weights. Response to the other treatments did not differ from that to the control in any vineyard. Crop reduction often leads to larger vines in the following year (Edson et al. 1993). The yield-to-pruning-weight ratio was highest in vineyard R1, followed by vineyards R2, H1, and H2. Only vineyard H1 fell into the optimal range of 5 to 10 (Smart and Robinson 1991), whereas vineyards R1 and R2 would be classified as overcropped and vineyard H2 as undercropped. The postharvest treatments did not modify the yield-to-pruning-weight ratio in any vineyard in either of the two years. This proportional change in yield and pruning weight reacted most strongly to manipulations of the postharvest period in the overcropped vineyards, and least in the undercropped vineyard. The difference in magnitude of the reaction in the two regions might be due to differences in vine reserve dynamics. The high crop loads may have reduced vine reserve accumulation until harvest (Candolfi-Vasconcelos et al. 1994), while in the low-yielding vineyards impact on reserves may have been minimal.

The impact of cropping levels on the vines was also revealed by the sugar yields per vine. In the first year, only vineyard R1 had a reduction in sugar yield per vine when leaves had been removed at harvest in the preceding season. In the second year, both Riverina vineyards and Hilltops vineyard H1 responded with increased sugar yield when the crop had been removed early in the previous two years and with decreased sugar yield when the leaves had been removed at harvest. The relative differences of these changes in sugar yield were not as great as the yield differences caused by the treatments, likely because fruit from the higher cropped vines was less mature at harvest than that from the lower cropped vines. A delay in fruit maturation is often associated with high crop loads (Koblet et al. 1996).

Trunk size and cane number and size.

Despite the differences in productivity levels and vine age, trunk girth was similar in all vineyards in the first season (Table 3⇓). In the second season, trunk girth of the more productive vines in the Riverina was larger compared to the first season, while in the Hilltops vineyards, the trunks were slightly smaller than in the previous year. The apparent decrease in trunk circumference in the Hilltops vineyards might have been due to poor growing conditions caused by drought. The cane numbers per vine were similar among the Riverina vineyards and vineyard H1 in the second season, but vines in H2 had almost 50% more canes (Table 3⇓). The impact of the treatments on cane numbers was inconsistent among vineyards and years, but indicates some impact on budburst, since the vines were pruned to similar bud numbers the previous season. For instance, in the first season (2001/2002) some treatments reduced cane numbers in vineyard R2, while in the following season they had no effect. In vineyard H2, crop removal led to the highest number of canes in both seasons, which may have been due to improved vine reserve status (Miller et al. 1997). With the exception of vineyard H2, differences in cane weight in winter (vigor) were more pronounced in the second season (2002/2003). In vineyard R1, cane weight was more than 60% lower in response to the leaf removal than to the crop removal after two years of implementation. Cane weight was more responsive than cane length (data not shown) to postharvest manipulations; in most cases the canes were lighter following leaf removal or leaf damage and heavier following crop removal or juice spray. In addition, the decrease in cane weight following leaf removal at harvest was stronger in the second season, suggesting that vine reserves were further depleted over consecutive seasons. These results also show that the differences in pruning weights were primarily due to cane size (diameter and length) rather than to cane number.

Yield components.

Clusters per vine and cluster weight were highest in R1, followed by R2, and were lowest in the Hilltops vineyards (Table 4⇓). Berries per cluster were similar in all vineyards, except in H2 where clusters had approximately one-third as many berries. In the first year, treatments affected yield components only in R1, where juice spray led to more berries per cluster. The yield reduction in response to leaf removal observed in R1 was attributed to fewer and smaller clusters. There was a trend in the Riverina, particularly in R1, but not in the less-fruitful Hilltops vineyards for leaf removal and leaf damage to decrease bud fruitfulness (clusters per shoot) (Table 3⇑). Crop removal produced about double the berries per bunch than the leaf removal in H1. In the Riverina vineyards, the treatments modified all yield components except berry weight. Leaf removal generally decreased and crop removal increased cluster and berry numbers and cluster weight. In addition, cluster weight in R1 was also increased by the juice-spray treatment. These results suggest that differences in carbon assimilation during the postharvest period caused variation in flower numbers and fruit set. Moreover, the differences in the number of clusters per shoot in the two Riverina vineyards indicate that the treatments also affected cluster initiation and/or differentiation (and hence bud fruitfulness) in these high-yielding vineyards. Vineyard productivity level changed with berry weight, which was highest in R1 and lowest in H2. However, the postharvest treatments failed to alter berry weights.

The potential crop produced by individual vines is comprised of several yield components determined at least 15 months before harvest (May 2004). The differences in yield components could be caused by altered reserve levels as indicated in spurs at leaf fall (Holzapfel and Smith 2005). Vines appear to be particularly susceptible to the influence of reserves during inflorescence initiation, differentiation, and bloom. Other studies have shown that both fruitfulness (cluster initiation; Keller and Koblet 1995) and fruit set (Zapata et al. 2004) are dependent on carbohydrate supply from current photosynthesis or reserve pools. Nevertheless, crop load and defoliation are generally thought to have little effect on cluster initiation and differentiation in the current season (Goffinet 2004). Reproductive development is most dependent on reserves during the budbreak period, when inflorescences must compete with unfolding leaves for access to reserves (Goffinet 2004). Therefore, cluster and flower numbers would be expected to be most vulnerable to poor reserve status during this period. However, Goffinet (2004) found no effect on flower initiation and development in Concord vines subjected to carbohydrate reserve depletion by defoliation and high crop load in the previous season, but fruit set was not reported. These observations indicate that changes in yield were attributable to berry numbers in the first season and also to bud fruitfulness in the second season.

Fruit composition.

Berry maturity at harvest (Table 5⇓) reflected the yield level, yield-to-pruning-weight ratio, and sugar yield reported above. At first sight, the cropping level in vineyard R1 might have been excessive and hindered fruit ripening, since vines with high crop loads often show delayed maturation (Koblet et al. 1996). However, although the crop load (yield-to-pruning-weight ratio) in vineyard R2 was only slightly lower than that in vineyard R1, fruit from those vines in both years ripened to commercial standards. Moreover, the concentration of sugar accumulated per berry was the highest in vineyard R1, suggesting that the low Brix levels were due to the large berry size achieved in this vineyard, perhaps the result of overirrigation. Indeed, this vineyard received the most irrigation (Table 1⇑), with the aim to produce high yields (Table 2⇑). While the treatments had little effect in the Hilltops vineyards, soluble solids in the Riverina vineyards was decreased by crop removal and increased by leaf removal after two seasons. Juice acid composition was also altered (Table 5⇓). Leaf removal resulted in lower TA and higher pH and crop removal had the opposite effect. Therefore, fruit maturity at harvest mirrored the crop level but not the crop load, suggesting that high yield rather than crop load was associated with delayed ripening (Jackson and Lombard 1993). This suggests an indirect effect of the postharvest period on berry composition in the following year, brought about by impacts on amount of fruit per vine.

The above conclusion is supported by the negative correlations between yield and soluble solids in each vineyard (data not shown), suggesting that higher crop levels decreased the rate of fruit ripening. However, a third factor may have also played a role. Intriguingly, there was no difference in vine balance (as indicated by the yield-to-pruning-weight ratio), which is generally thought to be more important than crop level for fruit quality (Kliewer and Dokoozlian 2005). Indeed, high yields were also associated with vigorous vegetative growth. These vines may have invested more resources in vegetative and reproductive growth rather than in storage reserves of the perennial parts of the vine. Conversely, smaller vines may have less sink demand for vegetative growth, which would allow more carbohydrates to be diverted to the fruit during the ripening phase. These vines could have a more effective leaf area for assimilate production because of better leaf exposure. Since all treatment plots of vines within the same vineyard were harvested on the same day, we do not know whether delaying the harvest date could have compensated for the slower ripening rates. However, in 2002 R1 was harvested much earlier than the other vineyards because of disease pressure and commercial demand for the grapes; consequently, R1 had a much longer postharvest period for the treatments to be effective.

The differing response to the treatments among vineyards, particularly between regions, could be due to a number of factors. First, production levels are much higher in the Riverina, suggesting that nutrient uptake and reserve accumulation during postharvest may be more important in sustaining the high yields. High crop level may cause low vine reserve status at harvest either because reserves are required to assist ripening the crop or because the vine cannot produce sufficient assimilates to replenish the storage reserves during the grape ripening period, or both. It has been suggested that high crop levels reduce the amount of vine reserves accumulated until harvest (Koblet et al. 1996) and that mobilization of reserves can occur if the supply of photoassimilate to the fruit is insufficient because of stress situations (Candolfi-Vasconcelos et al. 1994). Vines in the Riverina vineyards were not only more productive but also considerably younger (R1 was only in its third year at the start of the study) than those in the Hilltops vineyards, which probably affected their ability to manage with defoliation and cropping stress.

Second, averaged over the two treatment seasons (2000/2001 and 2001/2002) the Riverina vineyards received more water from irrigation and rainfall after harvest (+186 mm) than the Hilltops vineyards and more heat accumulation (+369 GDD) during the postharvest period. Therefore, vines growing in the Riverina are more able to produce assimilates and take up nutrients late in the season than vines growing in the Hilltops. Third, the drought during the 2002/2003 season, which primarily affected (nonirrigated) H2, probably masked some of the treatment effects.