1. Introduction
In recent years, Mediterranean regions are being affected by marked climate changes, primarily characterized by reduced precipitation, greater concurrence of temperature extremes and drought during the growing season, and increased inter-annual variability in temperatures and rainfall [
1,
2]. For this reason, the wine regions of southern Europe will experience, or are already experiencing, a modification of their traditional terroirs, which in turn, will significantly increase the variability of yields and wine quality attributes, style, and typicality [
3].
Soil is an important part of terroir, but great wines are not related to one soil type [
4]. Great wines are produced worldwide on a wide variety of soils and it is not possible to define a high-quality potential vineyard soil in terms of soil texture, soil type, or soil minerals [
5]. Soil depth, soil water holding capacity, and soil nitrogen level are important characteristics influencing vine behavior in terms of vigor, quantity, and quality of grapes. In Bordeaux (France), the best soils are considered those characterized by free draining, lack of water logging in the rooting zone, and limited water availability during ripening [
6]. Indeed, some researchers give more importance to soil physical properties determining water supply to the vine rather than to soil chemical constituents [
7]. Whether soil plays a primary and direct role on wine quality or it indirectly determines wine quality, through its effects on vine growth, canopy density, and vigor must be still established [
8].
Soil texture influences soil water holding capacity and consequently the progressive water release for vine root uptake. Regular but limited water supply before veraison reduces shoot and berry growth and increases berry anthocyanin and tannin concentration [
4]. van Leeuwen et al. [
9] reported that different non-irrigated cultivars grown on clay soil displayed a higher level of sugar and lower total acidity compared to those on gravel and sandy soils. However, site and vintage seem to play major roles (rather than soil texture) on vine size and wine sensory characteristics [
8].
Generally, high-quality red wines need moderate water deficit and mineral supply, especially nitrogen [
4]. Hence, irrigation may be needed to avoid severe vine water stress occurring in some vintages and in soils characterized by low holding capacity. In recent times, irrigation is becoming a common practice also to prevent the negative effects of climate change [
10,
11] even in those areas where grapevines are traditionally rain-fed. Indeed, irrigation allows to standardize yield and quality of grapes over the years, especially when rainfall is too low and, in some circumstances, certain irrigation techniques allow for significant water savings [
3].
Nevertheless, water-deficit-induced different effects on grapevine yield and quality components depend on timing, intensity, and duration of plant water stress [
12]. The adoption of regulated deficit irrigation (RDI) in the vineyard requires setting irrigation timing and volume to keep the vines within the stress level ranges identified for the different phenological phases. Generally, moderate-to-high water stress imposed with RDI positively influenced the amount of berry secondary metabolite, including total anthocyanin content [
13,
14], and induced a reduction in berry size [
15,
16,
17]. However, in the areas with rainy spring and/or high soil water reserves, irrigation may be used to modulate vine water stress levels only in mid to late summer [
18].
In deficit irrigation strategies, it is useful to define irrigation volume based on fractions of crop evapotranspiration (ET
c) when crop coefficients are well adjusted and the reference evapotranspiration (ET
0) is available [
19]. This, however, may create some indecision as plant water stress development depends not only on the fraction of water consumption replaced in the soil, but also on soil water holding capacity, on growing conditions, and so on [
20]. Different methods to evaluate the vine’s water status have been implemented and are available, with the advantage of possible automation [
21,
22]. One alternative approach is the use of plant water potential as an irrigation signal. Predawn and stem leaf water potentials are the most-used methods for the current irrigation management of vineyards. Although they are considered good indicators of vine water status [
23], not all researchers consider them equally correlated with soil water content [
24]. Predawn leaf water potential, in particular, is thought to estimate soil moisture better than the other methods [
23,
25,
26] because it is nearly not affected by environmental factors, representing a situation of stable equilibrium between soil and atmosphere. On the contrary, other authors [
27,
28] showed that midday stem water potential is a very accurate measure and a better indicator of grapevine water status than predawn leaf water potential for the irrigation management of grapevines. These different results may be in part explained by the fact that grapevine cultivars withstand different levels of drought as they can show similar values of stem water potential, but at the same time, different values of predawn water potential [
29]. For this reason, it is difficult to establish a range of stem water potential to manage vineyard irrigation that would be valid for all varieties. Hence, in this experiment, we decided to use predawn leaf water potential to establish irrigation timing.
Nero d’Avola (also known as Calabrese) is characterized by high vigor and yield [
30] and is the main black cultivar in Sicily, accounting for about 15,500 ha (15.6% of total Sicilian vineyards) [
31]. ‘Nero d’Avola’ is in 17 Sicilian PDO wines with a production of 31 mL out of 150 mL of total Sicilian PDO wines [
32]. Recently, ‘Nero d’Avola’ has been planted in other parts of the world including Australia, California, and South Africa, also thanks to its ability to withstand dry, hot conditions. In one of the few studies carried out on ‘Nero d’Avola’, the vines showed a rather isohydric behavior, its degree varying with potassium availability when vines were subjected to moderate drought stress [
33]. Nevertheless, the real ability of ‘Nero d’Avola’ to adapt to drought conditions in open field must be still determined.
The trial was carried out in Sicily, during two vegetative seasons (2005 and 2006) and in a ‘Nero d’Avola’ vineyard characterized by two soils different for their position in the vineyard (one at the top, the other one at the bottom of a slope), depth, water holding capacity, organic matter, and nitrogen content. The objectives of the present study were to test (1) the effect of soil and RDI irrigation on vigor, yield, and quality of ‘Nero d’Avola’ grapes; and (2) the ability of ‘Nero d’Avola’ vines to adapt to drought conditions in open field.
2. Materials and Methods
2.1. Experimental Site
The experimental site was located within the Alcamo D.O.C. area, in the hinterland of western Sicily (37°55′11.66″ N; 13°04′10.03″ E), 300 m a.s.l. The trial was carried out for two years (2005–2006) in a drip-irrigated vineyard with eight-year-old ‘Nero d’Avola’ vines grafted onto 1103 P rootstock. The rows were spaced 2.40 m apart, with 0.95 m between vines on the row. Vines were trained to a vertical shoot positioned trellis (VSP) and pruned to two buds per spur, spaced at approximately 15 cm in a single cordon (
Table S1). Climate trends of the two years of study are perfectly in line with climate trends of recent years (2015–2019, data not shown); therefore, data collected in 2005 and 2006 can be considered relevant and representative of the present climatic conditions at the experimental site.
The irrigation system used 4 L h−1 pressure-compensating inline emitters spaced 0.95 m apart. Drip-irrigation lines were placed ~40 cm above ground. The amount of water applied in each irrigation treatment was measured using flow meters.
Soil management practices consisted of growing a cover crop (Vicia faba) during winter and burying its biomass in the soil in April. Three shallow tillage events (10–12 cm deep), from spring to summer, were carried out to control weeds, prevent crust formation and ultimately reduce soil evaporation.
Weather data were collected with an automated weather station located at 680 m of distance (37°45′50.61″ N, 13°04′13.73″ E) from the experimental vineyard.
2.2. Soil Characteristics
Analysis of 20 profiles allowed for the individuation of two distinct types of soil, one at the top, the other one at the bottom of the hill (slope 10%). The two clay soils were classified using the soil taxonomy [
34] as Typic Xerorthents (entisol) at the top of the hill, and Chromic Haploxererts (vertisol) at the bottom of the hill. The entisol was characterized by a loamy texture and clay-montmorillonite with a low amount of skeleton and pH of 7.8. The entisol profile was Ap-C and its depth was up to 50–60 cm. Organic matter (0.98%), total salinity (0.98 mS/cm), CaCO3 (5.4%) and total N content (658 ppm) were low, while exchangeable K
2O was high (294 ppm).
The texture of vertisol was characterized by 41.3% of clay; the profile was very deep (>100 cm) and the amount of skeleton was generally low. The pH was 7.6 and the organic matter was high (2.03%). Total salinity (1.07 mS/cm) and CaCO3 content (5.1%) were low. Exchangeable K
2O was high (402 ppm), while total N was medium/low (966 ppm). When not cultivated during the dry season, both soils exhibited deep cracks (to a depth of 80–100 cm in the vertisol) (
Table S1).
2.3. Irrigation Treatments
In 2005 and 2006, the following treatments were applied:
(1) rain-fed—vines grown without irrigation water;
(2) regulated deficit irrigation (RDI)—vines irrigated at 25% of estimated ET
c up to the end of veraison and at 10% of estimated ET
c up to 10 days before harvest. Predawn leaf water potential (PDWP) was used to establish irrigation timing according to specific thresholds differing by phenological stages. In particular, the irrigation was applied when PDWP reached values below –0.5 MPa until the end of veraison, and below –0.7 MPa after the end of veraison until 10 days before harvest (
Table S1). ET
c was calculated with the method proposed by Allen et al. [
19] using the Penman–Monteith equation to calculate reference evapotranspiration (ET
0) and tabulated crop coefficients (Kc). The rain-fed treatment was considered the reference (control) as this is the common irrigation management in the area, and a full irrigated reference (irrigation = 100% ET
c) would be meaningless for the production of quality wines.
2.4. Experimental Design
The experiment layout was a split-plot design with the two soils as the main plots and the two irrigation treatments as sub-plots. The experimental plot was first divided in six main-plots of six rows each for the assignment of the entisol and vertisol. Each main-plot was further divided in two sub-plots for the assignment of the two irrigation treatments. The elemental plot comprised three adjacent rows (two buffer rows and a central one for data collection) and was replicated three times. Ten vines with similar vigor, measured at the beginning of the trial, for each replicate were considered for data collection (
Figure S1).
2.5. Soil Water Content
Soil water content was monitored with one Diviner 2000 capacitance probe (Sentek Environmental Technologies, Stepney, South Australia) per treatment and soil, each placed in access tubes installed to a depth of 1 m and located in the row line at 47 cm from the trunk in all treatments. Moreover, field capacity and wilting point were determined using Richard’s plates [
35] to calibrate the probe with the vineyard’s soil type. Data from the probes were expressed as a percentage of volumetric water content. Measurements were carried out every 10 days starting about 1 month after fruit set. For each, sampling date, treatment and soil, and soil water content measurements (six) were taken at 15 cm to 100 cm depths (at about each 15 cm).
2.6. Ecophysilogical Measurements
The measurements of PDWP were taken before sunrise, with stem water potential (SWP) measurements taken at midday (between 12:30 to 13:30 h), using a Scholander pressure chamber [
36] and selecting four leaves, from four different vines and opposite to clusters for each treatment and replication. Following the methodology described by Williams and Araujo [
23], SWP was measured on mature leaves with similar age in the shaded canopy side of the vertical trellis. At least 60 min before, the leaves were enclosed in plastic bags and covered with aluminum foil [
23]. Four measurements per replication of PDWP and SWP were collected starting at fruit set every 10 days in each year.
Stomatal conductance (gs) (four per replication) was measured at midday in the same days and on similar leaves (but exposed to the sun) as those of water potentials, using an AP4 porometer (Delta-T Devices Ltd., 130 Low Road, Burwell, Cambridge, UK).
2.7. Leaf Area and Pruning Mass
At two phenological stages (pea size and harvest), 10 shoots per treatment and replication were collected, and total leaf area (TLA) of primary and lateral shoots was measured using an LI-3100C area meter (Li-COR Environmental, 4647 Superior Street Lincoln, NE, USA). During winter of both years, pruning wood of ten vines for each replication and treatment was weighed and cane number per vine was counted.
2.8. Yield and Grape Composition
At full maturity, the clusters of 10 vines per treatment and replicate were harvested, counted and weighed and then used to calculate yield per vine and average cluster weight. For each replication and treatment, 100 berries were randomly collected and weighed, and average berry weight was calculated. Total anthocyanins (expressed as mg kg
−1 of grapes and mg/berry) [
37] were measured from a sample of 25 berries per replication and treatment. The berries were randomly collected from different sections of the bunch (top, middle, bottom, inner and outer portions). The skins of each sample were separated from the pulp and placed in a flask containing 25 mL of tartaric buffer (500 mL of distilled water, 5 g of tartaric acid, 22 mL of 1 N NaOH, 2 g of sodium metabisulphite and 120 mL of 95% ethanol; pH 3.2). The buffer volume was adjusted to 1 L with distilled water. Skins were placed in the buffer for 4 h at room temperature prior to homogenization and centrifugation. The supernatant was collected in a 100 mL volumetric flask. The residue was washed again with tartaric buffer, added to the volumetric flask, and the volume was raised to 100 mL with the same buffer. The extract (10 mL) was diluted 25 times with acidified ethanol (ethanol, H2O and concentrated HCl, 70:30:1
v/v/v), and the absorbance was read at 540 nm using a UV–vis spectrophotometer (Varian Cary 50 Bio UV–vis Spectrophotometer, McKinley Scientific, Sparta, NJ, USA).
For each replication and treatment, bunches were hand pressed, and the juice was recovered to measure total soluble solids (TSS, °Brix) using an Atago PR-32 digital refractometer (Atago, Tokyo, Japan) and titratable acidity (TA) using a Crison Compact Titrator (Crison Instruments, Barcelona, Spain) by raising pH to 7 with 0.1 N NaOH (expressed in g L−1 of tartaric acid).
2.9. Statistical Analysis
Productive, qualitative, vegetative, and soil water content data were processed by a two-way mixed model analysis of variance (ANOVA) to evaluate the effects of the main factors (soil and irrigation) and their possible interaction. Year was considered as a random factor.
In each season, PDWP and SWP data were processed by repeated measures two-way ANOVA (main factors: soil and irrigation). Then for each date, Tukey’s multiple comparison test was used to detect different means. Moreover, in each season, the relationships between predawn water potential (PDWP) and stem water potential (SWP) and between these parameters and stomatal conductance (gs) were tested using regression analysis. Difference between slopes were tested by the Student’s t-test. Log-transformation of data was performed, when necessary.
All statistical procedures were performed using R (R Core Team, 2020) [
38]. Mixed model ANOVA was performed using the ‘lme4’ R library [
39].