Abstract
Manual horizontal cluster division may constitute an efficient tool to optimize wine composition. To test this hypothesis and to determine the optimal timing of this viticultural practice, the impact of cluster division on cluster morphology, bunch rot disease severity, and harvest parameters such as yield and total soluble solids was examined on the white Vitis vinifera L. cultivars Pinot gris and Riesling in the Luxembourgish winegrowing region in 2010 and 2011. Treatments were applied at five different times between prebloom (BBCH 57) and veraison (BBCH 81). In both varieties, all postbloom cluster division treatments were able to loosen the cluster structure and reduce the bunch rot disease severity. The later cluster division took place, the more pronounced were the effects on cluster structure. The density index was a suitable measure of the cluster structure and the predisposition to bunch rot infestation. Cluster division resulted in yield reductions (up to 30%), which increased with time after flowering. Postbloom cluster division may help to optimize wine quality by improving grape maturity due to crop load reduction, reducing fungal contamination, and improving grape composition through the possibility of elongating the ripening period. Postbloom horizontal cluster division can be recommended as an efficient crop cultural tool for premium wine production under cool-climate conditions.
- Botrytis cinerea
- bunch rot
- cluster structure
- crop cultural practice
- density index
- grape maturity
- viticulture
- Vitis vinifera
Grape bunch rot or gray mold caused by the fungus Botrytis cinerea leads to economic damage worldwide. In addition to yield loss, B. cinerea infection can result in decreased wine quality by the generation of off-flavors, unstable color, oxidative damage, premature aging, and difficulties in clarification (Ribéreau-Gayon 1983, Smart and Robinson 1991). Moreover, B. cinerea infection may favor infections by other fungi and bacteria, reinforcing the occurrence of wine defects such as earthy, moldy off-flavors (La Guerche et al. 2005, Smart and Robinson 1991). Berry infection by B. cinerea at an early stage of ripening reduces potential wine quality because of berry decay (Kassemeyer and Berkelmann-Löhnertz 2009) and obstruction of the maturation process. Whereas, under specific conditions, gray mold infections of fully ripe berries may even improve grape composition (Kassemeyer and Berkelmann-Löhnertz 2009) due to the concentration of grape ingredients. However, the aim of any bunch rot protection strategy is to avoid infections on unripe berries.
In compact, dense grape clusters the conditions are generally favorable for infection and spread of fungal pathogens such as B. cinerea due to a combination of three effects. First, in the interior parts of dense clusters, the air circulation and sun exposure are low (English et al. 1989), resulting in slow drying processes, high interior humidity, and, consequently, favorable conditions for the establishment of fungal pathogens (Zoecklein et al. 1992). Second, the dense structure further reinforces the risk of berry burst induced by berries touching each other because of high internal pressure (Smart and Robinson 1991). Burst berries are easy for fungal pathogens to colonize (Evans 2008) and may represent starting points for rapid spread of bunch rot. Third, the close proximity of the single berries in compact clusters allows for faster berry-to-berry spread of fungal pathogens. Further, results of Percival et al. (1993) indicate that cuticles are thinner on berries touching each other, and according to Gabler et al. (2003), cuticle and wax contents are negatively correlated with susceptibility to B. cinerea.
Several studies demonstrate the strong link between the compactness of grape clusters and their predisposition to severe bunch rot infestation (Evers et al. 2010, Hed et al. 2009, 2011, Molitor et al. 2011a). Thus, all treatments that reduce cluster compactness are expected to lead to a lower predisposition of clusters to early and severe bunch rot incidence.
Loosening of clusters can be induced either by application of chemical compounds such as bioregulators or through crop cultural practices. For example, the application of the bioregulators Regalis (Lo Giudice et al. 2004, Molitor et al. 2011b) or Gibb3 (Evers et al. 2010, Hed et al. 2011) in the period around grape flowering reduced the compactness of clusters due to a reduction in berry number per cluster or/and berry size. However, the effects were dependent on variety, application stage and dose, and weather conditions. Further, some authors describe unacceptable crop losses in the present (Lo Giudice et al. 2004) or in the subsequent season (Weyand and Schultz 2005) upon the application of bioregulators. Another approach to loosen grape clusters is to elongate cluster stems. Applications of gibberellic acid between budburst and beginning of flowering were able to induce such an elongation (Weaver 1975). However, Molitor et al. (2012) showed that crop losses might be possible in the season after the application. Consistently positive effects on cluster structure and bunch rot infestation were observed as a result of a leaf removal in the cluster zone between shortly before flowering and the beginning of bunch closure (Molitor et al. 2011a). Thus, cluster-zone leaf removal has become a widely used crop cultural method in many winegrowing regions (Molitor et al. 2011a, Zoecklein et al. 1992).
In the past, a major focus of many grapegrowers was to maximize grape yield and viticultural practices resulting in a yield reduction were of minor interest. In the last decades, many growers of high-quality grapes have focused on methods to regulate crop load (Hed et al. 2011). However, cluster thinning (removal of whole clusters) usually leads to a compensation reaction of the plant: the remaining clusters are better supplied by assimilates, berries per cluster and berry weights increase (Naor et al. 2002), and clusters become more compact, which favors bunch rot incidence (Becker 2011, Schultz et al. 2003). However, both healthy and fully mature grapes and moderate yields are required for the production of high-quality wines. In this sense, a manual horizontal division of compact grape clusters might be an innovative option to maximize wine quality by combining a regulation of the crop load and a reduction of cluster compactness (i.e., predisposition to bunch rot incidence). Initial trials with these new crop cultural techniques were focused on either crop-load reduction (Fader et al. 2004, Fox 2005, Schultz et al. 2003) or bunch rot control (Hed and Travis 2005, 2006, 2007). However, systematic studies on both aspects are needed. Further, little is known about the optimal timing of manual horizontal division of compact grape clusters or its applicability to different varieties.
Thus, to confirm the hypothesis that cluster division could be an efficient tool to optimize berry composition and to investigate the optimal timing of this viticultural practice, the present study on the Vitis vinifera L. cultivars Pinot gris and Riesling focused on the impact of the timing of cluster division on cluster morphology, bunch rot epidemic, and harvest parameters such as yield and total soluble solids.
Materials and Methods
Vineyard site and experimental design.
The study was carried out in 2010 and 2011 in the experimental vineyards of the Institut Viti-Vinicole in Remich, Luxembourg (49.54 N, 6.35 E) in the Moselle Valley on the white Vitis vinifera L. varieties Pinot gris and Riesling. These varieties were chosen because they are widely grown in the region and susceptible to bunch rot due to their usually compact cluster structure. The vineyards were planted in 2000 and the vines grafted onto SO4 rootstock were trained to a vertical shoot-positioning system with two canes per vine.
Cultural management was identical for all treatments. To protect the vineyards, background applications against Plasmopara viticola and Erysiphe necator were carried out during the whole season. No fungicides with activity against B. cinerea were applied.
The field experiments used a randomized block design with four replicates of eight vines per treatment. Treatments were the same in all four experiments (two varieties, two years) and defined as follows: 0 (untreated control, no cluster division), 57 (inflorescence clipping at BBCH 57) (Lorenz et al. 1995), 73 (cluster stripping at BBCH 73), 77 (cluster division at BBCH 77), 79 (cluster division at BBCH 79), and 81 (cluster division at BBCH 81) (Table 1).
In the cluster division treatments, all berries in the lower half of the clusters were removed. In treatment BBCH 57, the lower parts of the inflorescences were clipped off manually. The cluster stripping at BBCH 73 consisted in a manual removal of the berries on the lower part of the clusters by pulling them through closed hands. Here, the stems remained on the cluster. Horizontal cluster division (in the middle of the stems) in BBCH 77, 79, and 81 was performed using vineyard scissors.
Determination of cluster structure.
To study the impact of the treatments on cluster structure, an assessment of the density index was conducted as described elsewhere (Evers et al. 2010, Molitor et al. 2011b). The density index classifies the cluster structure between 1 (very flexible) and 5 (very compact) and was assessed by examining 100 clusters (50 on each side of the row) per plot (six treatments with four replicates each) at the phenological growth stage BBCH 79 (11 Aug 2010; 20 Jul 2011 for Pinot gris and 27 Jul 2011 for Riesling).
In addition, 10 randomly selected clusters were collected in each plot after veraison (2 Sep 2010; 22 Aug 2011 for Pinot gris and 2 Sep 2011 for Riesling) to determine cluster weight (with and without stems) and berry number. The average berry weight was calculated by dividing the cluster weight without stems by the number of berries.
Assessment of B. cinerea disease severity.
Disease severity of B. cinerea was assessed in the field three (Pinot gris 2011) to six times (Riesling 2011) at weekly or twice-weekly intervals between veraison and harvest by examining 100 randomly selected clusters per plot (50 on each side of the row) according to the EPPO guideline PP1/17 classifying visually observed disease severities in seven classes (0%, 1–5%, 6–10%, 11–25%, 26–50%, 51–75%, 76–100%). The average disease severity was calculated by summing up the number of observations per class multiplied by the midvalue of the class interval and dividing this sum by the total number of observations (100).
To describe the temporal progression of the disease severity, the average values were plotted against the date of assessment (expressed as days after bloom; BBCH 68) for every treatment. Progress curves were fitted to these data according to the exponential equation y = ea(x − xo), where y = average disease severity, a = slope of the curve, x = days after bloom (BBCH 68), and xo = days after bloom (BBCH 68) reaching a disease severity of 1% as described previously (Evers et al. 2010, Molitor et al. 2011b). This allows estimating the time at which a certain disease severity value would be reached. In the present examinations, the xo - values (days after bloom [BBCH 68] reaching a disease severity of 1%) were selected to compare the disease progress in the different treatments. Subtracting the xo - values of the treatments from the xo - values of the control allows to estimate the temporal displacement of the day after bloom (BBCH 68) at which a treatment reached a disease severity of 1%.
Determination of total soluble solids and yield.
Close to harvest, 100 berries per plot (50 on each side of the row) were randomly sampled. After pressing, the juice was centrifuged and the total soluble solids were measured by Fourier transform infrared (FOSS NIRSystems, Laurel, MD). Each plot was harvested separately. The average yield per plant was calculated by dividing the yield recorded per plot by the number of plants.
Meteorological data and statistical analysis.
Daily average temperatures as well as daily precipitation sums were recorded directly at the experimental site (Figure 1). Despite the different division techniques, the direct impact on the cluster morphology was comparable at all division moments. Hence, parametric data were analyzed for the effect of the timing of the cluster division by one-way ANOVA. For multiple comparison procedures between means, Tukey tests were performed using PASW Statistics 18. Further regression analyses were conducted with the same program. Ordinal data sets (density index) were analyzed with Kruskal-Wallis one-way ANOVA. If null hypothesis was rejected, pairwise multiple comparisons were conducted using IBM SPSS Statistics 19. P values were adjusted according to Hochberg (1988).
Results
Cluster morphology.
Inflorescence clipping did not significantly reduce the density index in any trial (Table 2). Cluster stripping at BBCH 73 numerically decreased the density index in all trials, although not by a statistically significant amount. Cluster divisions at BBCH 77 of Riesling in 2010 and Pinot gris in 2011 led to significant reductions of the density index as compared to the control. The same was the case in all trials for the cluster division at BBCH 79. With the exception of Riesling 2010, the value of the density index constantly decreased with later cluster division (except for BBCH 81, where the assessment of the density index occurred before cluster division).
Cluster division after flowering significantly reduced cluster weight and berries per cluster when compared to the control. Early inflorescence clipping at BBCH 57 significantly reduced cluster weight for Pinot gris in 2011 and berries per cluster for both varieties in 2010 and Pinot gris in 2011. In Riesling, the number of berries in BBCH 57 was in both years statistically higher than in all treatments with a cluster division after flowering. With the exception of Riesling in 2011, cluster division numerically increased single berry weights when compared to the control, but the increase was significant only in cluster division at BBCH 73 and 77 in Pinot gris 2011 (Table 2).
Disease severity.
In Pinot gris, all division treatments significantly reduced the final disease severity as compared to the control (Table 3). In Riesling, the divisions at BBCH 77 and 79 led to a statistically significant reduction of the final disease severity. If cluster division was conducted between flowering and veraison, an efficacy (1 - relative disease severity) of at least 45% was reached. The highest average efficacy was recorded in BBCH 79 (73.8%). A strong correlation (except with BBCH 81) between the density index and the final disease severity of B. cinerea was observed. An exponential function of the type y = ea(x − xo) was well adapted to the present data (Pinot gris 2010: R2 = 0.96, p = 0.003; Riesling 2010: R2 = 0.77, p = 0.052; Pinot gris 2011: R2 = 0.99, p < 0.001; Riesling 2011: R2 = 0.93, p = 0.008) (Figure 2).
Disease progress.
The exponential equation (y = ea(x − xo)) was well adapted to the observed disease progress. The coefficient of determination (R2) ranged in all cases above 0.95 and the p values below 0.05 (exception, Pinot gris 2011) (Figure 3; Table 4). According to the proposed formula, all treatments postponed the time of reaching 1% disease severity compared to the untreated control except the inflorescence clipping at BBCH 57 in Riesling 2011. The delay ranged on average of all trials between 5.6 (BBCH 57) and 28.2 days (BBCH 79) (Table 4).
Harvest parameters.
Cluster division reduced the final yield at harvest. In both years, the reduction was statistically significant in Riesling if cluster division occurred postbloom (BBCH 73 to 81), with the exception of BBCH 73 in 2011, whereas in Pinot gris, that was the case just for the late division dates BBCH 79 and 81 in 2011 (Table 3). With the exception of Pinot gris in 2010, yield constantly decreased with later cluster division dates. Prebloom cluster division reduced the yield on average by only 8%, whereas the yield reduction in the late treatment at veraison reached on average 30.4%.
In all trials but Riesling 2010, where total soluble solids were almost identical in all treatments, the total soluble solids at harvest were increased by all cluster division treatments. The highest total soluble solids was always measured in the treatment with the latest cluster division (BBCH 81). With the exception of Riesling in 2010, cluster division at BBCH 79 or 81 significantly increased the total soluble solids in all trials (Table 3). In three trials (exception: Riesling 2010), there was a statistically significant negative correlation between yield and total soluble solids (R2 = 0.76, Pinot gris 2010; R2 = 0.83, Pinot gris 2011; R2 = 0.92, Riesling 2011; p < 0.05, each).
Discussion
Effects on cluster structure.
Cluster division after flowering (between BBCH 73 and 79) loosened the cluster structure. This effect was caused by a reduced number of berries per cluster and by a rearrangement of cluster morphology. After the removal of the lower parts of the clusters, the remaining berries partly moved into the emerging gap, leading to a reduction in cluster compactness and in the number of berry-to-berry contacts. Due to this rearrangement, the central—and generally most compact—parts of the cluster were loosened. Overall, the loosening effect was stronger (as shown by lower density index values) the later the cluster division took place.
The loosening effects observed in the treatment with cluster division prebloom (inflorescence clipping at BBCH 57) were much lower than in the postbloom treatments and berry numbers as well as cluster weights were less reduced. In general, fruit set in grapes is determined by the supply of carbohydrates (Caspari et al. 1998) and the degree of abscission is related to the sugar concentration in the inflorescence shortly after anthesis (Vasconcelos et al. 2009). Since in the present examinations the assimilating leaf area was unaffected by the flower reduction and the number of single flowers per cluster was reduced due to inflorescence clipping, more assimilates were available per single inflorescence. We assume that, due to this, more berries and higher berry weights (of the remaining berries) were observed in the treatment with prebloom inflorescence clipping at BBCH 57 as compared to the postbloom treatments.
Reduction of bunch rot severity.
In contrast to classical cluster thinning (removal of complete clusters) (Schultz et al. 2003, Becker 2011), cluster division consistently reduced bunch rot disease severity. In general, the reduction of cluster compactness is expected to diminish the pressure inside the cluster, avoid berry burst, and allow a faster drying of formerly interior parts of the clusters due to the morphological rearrangement (Hed et al. 2009). Indeed, in the present trials, a lower final B. cinerea disease severity and clear delay of the date reaching 1% disease severity was induced by all postbloom cluster division treatments. In addition, final disease severity generally decreased the later cluster division was performed, confirming the recommendation that cluster division should be timed as late as possible to limit compensation effects (Fox 2005). In the present study, a strong correlation was observed between the reduction of the density index and bunch rot susceptibility, explaining 77 to 99% of the variation in the final disease severity in all treatments with cluster division before veraison.
In contrast to the earlier treatments, the efficacy of late cluster division at veraison was less consistent over two years and was presumably influenced by two factors: the degree of bunch rot already established in the interior parts of the cluster at the moment of division and the weather conditions in the following period. Late (at veraison) divided clusters grow under unfavorable morphological and microclimatic conditions throughout most of their development, meaning that early bunch rot could have already established in the interior parts of the clusters. Even if cluster division at veraison removes previously infected berries, some may remain and provide an inoculum source for a further spread within the clusters. Moreover, the injuries at single berries caused by the vineyard scissors could be a gateway for fungal pathogens (Kassemeyer and Berkelmann-Löhnertz 2009) and might be an additional source for a slightly higher infestation level in the treatment with cluster division at veraison. Berries wounded by scissors at their unripe stage prior to veraison were observed to dry out rapidly and were not colonized by B. cinerea (Fox 2005).
Existing practical recommendations for cluster division were mainly focused on the period just before or at bunch closure (Fader et al. 2004, Hed and Travis 2005, 2006, 2007, Schultz et al. 2003). The present examinations confirmed the adequacy of this period. However, since in most of the trials presented here the final disease severity was also reduced significantly as a result of the late cluster division at veraison, this treatment can be seen as an option if cluster division is not practicable during the earlier, labor-intensive period.
Even though cluster modification by manual cluster stripping of the berries at BBCH 73 was slightly less efficacious than the later treatments due to stronger compensation effects, the period in which cluster division may have a positive effect on the health status was quite long (fruit set to veraison). However, this result has to be confirmed under divergent climatic conditions. In the present trials, the efficacy to control bunch rot due to cluster division between fruit set and veraison (45 to 96%) was clearly higher than that reached by the use of botryticides or bioregulators in earlier trials (Evers et al. 2010, Molitor et al. 2011b). In general, the observed effects were aligned in both varieties. However, the benefits of the treatments increased with the compactness of clusters.
In the studies of Beresford et al. (2006), a log function proved to be adequate to describe the bunch rot epidemics. However, under the present experimental and climate conditions and the observed range of disease severities, an exponential equation (y = ea(x - xo)) was best adapted to the epidemics of B. cinerea, confirming other observations (Evers et al. 2010, Molitor et al. 2011b).
Influence on quantity and quality.
All postbloom cluster division treatments reduced yield as a consequence of the removal of the lower parts of the clusters; the reduction was more pronounced the later the cluster division occurred (less compensation). However, assuming that for premium wines heavily rotten grapes are discarded at harvest, the effective yield loss caused by postbloom cluster division could be considerably lower. Moreover, the culling process can be avoided or at least diminished due to a lower percentage of rotten grapes, which could compensate for at least part of the labor cost necessary for the cluster division, estimated at 75 to 100 hr/ha (Schultz et al. 2003, Fader et al. 2004).
Total soluble solids contents showed strong correlations to yields. According to the natural quantity–quality relation, reduced grape yields generally force higher total soluble solids content if the leaf area is constant (Kliewer and Dokoozlian 2005). In the present study, except for Riesling 2010, the yields explained 75.6 to 92% of the variance in total soluble solids. Higher total soluble solids content in the must is usually linked to greater maturity, at least under the cool-climate viticulture conditions in the Moselle Valley. Thus, the yield reduction caused by cluster division might allow for a potential improvement of wine quality.
The presence of B. cinerea, especially at early (unripe) stages of berry development, may negatively affect wine composition because of berry decay (Kassemeyer and Berkelmann-Löhnertz 2009) and hinder maturity development. Moreover, more thermophile secondary mold pathogens, such as Penicillium expansum or sour rot-inducing bacteria or yeasts, may increasingly appear if berries are rotten at an early (and normally warmer) stage. Consequently, the temporal delay of the Botrytis epidemic caused by postbloom cluster division and earlier maturation diminishes the risk of fungal contamination on unripe grapes.
In cool-climate winegrowing regions, the timing of the harvest date is generally determined more by crop health status than by optimal maturity. Practical observations indicate that a delay of harvest date due to a better crop health status is able to increase grape maturity and to decrease the risk of atypical aging flavors. Consequently, wines produced from late-harvested grapes are generally preferred in wine tastings (Spring 2004). As cluster division treatments postpone the epidemic, the date of Botrytis reaching a threshold of tolerable bunch rot disease severity is also delayed. Consequently, postbloom cluster division may enable an extension of the ripening period in the vineyard.
Conclusions
Susceptibility of grapes to bunch rot is to a major extent determined by cluster structure. Manual horizontal cluster division is able to reduce cluster compactness and the number of berry-to-berry contacts. The positive effects on cluster structure are more pronounced with later cluster division. According to the present study, postbloom horizontal cluster division can be recommended as an efficient crop cultural tool for premium winegrape production under the climatic conditions of the Luxemburgish winegrowing region by improving grape maturity due to crop-load reduction, reducing fungal contamination of grapes, and improving grape composition through the possibility of extending the ripening period.
Acknowledgments
Acknowledgments: The authors thank the Institut Viti-Vinicole (Remich, Luxembourg) for financial support.
The authors specially acknowledge B. Untereiner, S. Contal, C. Walczak, L. Solinhac, N. Kinlen, J. Koch, R. Mannes, S. Fischer, and C. Blum and the teams of the CRP-Gabriel Lippmann and the Institut Viti-Vinicole for their support in the experimental vineyards and in the laboratory and M. Beyer and T. Bohn for help with statistical analyses.
- Received March 2012.
- Revision received June 2012.
- Accepted July 2012.
- Published online December 2012
- ©2012 by the American Society for Enology and Viticulture