Introduction

Vineyard fertilization has been and still remains a common cultural practice carried out to achieve different objectives mainly related to grape quality. However, nutritional imbalances often occur together with a loss in must quality. For instance, the excessive intake of potassium promotes a loss of acidity which results in reduced colour stability and poor taste (Kodur, 2011). Grape quality and agronomic practices that respect the environment are top priorities in current viticulture versus high yield criteria. Yield and quality are closely linked to the nutritional status of the crop (Champagnol, 1990); however, the nutritional requirements leading to a quality vintage are not yet fully established and therefore continue to be a matter of great interest and ongoing research. In this sense, both plant tissue and soil analyses have been widely used to characterize the nutritional status of the vineyard (Kliewer, 1991; Robinson, 2005), and leaf analyses are widely recognized as the most reliable method of determining the nutritional status of grapevines (Lucena, 1997).

The most common methods for nutritional diagnosis of leaf tissues are the Critical Values and the Sufficiency Ranges (SR) methods. Both methods use the individual total nutrient concentrations in dry tissues separately, comparing them to reference values obtained from optimal populations (Sumner, 1978; Lucena, 1997).

The Deviation from Optimum Percentage (DOP) method is a routine analysis interpretation method which compares nutrient concentrations to the references using a percentage expression (Montañés et al., 1993; Lucena, 1997). This method quantifies the difference between a single nutrient concentration and its reference value, offering the advantage of ranking the order of requirements or order of limitation from the most negative to the highest positive nutrient index (Montañés et al., 1993). DOP indexes help us to determine which nutrients need to be included in a fertilization program (Montañés et al., 1993; Monge et al., 1995).

Furthermore, the sum of the absolute value of the different DOP indexes (∑│DOP_i│) is a general index which represents the complete nutritional balance of the plant and indicates the importance or severity of an anomalous situation. However, the DOP method is not widely used mainly due to the lack of useful references for many crops (Lucena, 1997).

A large data bank of leaf analysis gathered over a long period of time is necessary to establish adequate and representative references with the objective of monitoring the nutritional status of a vineyard within the environmental and cultural factors of a particular area (Sumner, 1978; Failla et al., 1993b; Lucena, 1997; García-Escudero et al., 2013). Large data banks are often used to define desirable ranges for leaf analysis interpretation (Sumner, 1978; Lucena, 1997; García-Escudero et al., 2013) through statistical operations that explain the variability of nutrient concentrations inside the population (Sumner, 1978). As a result, different sufficiency ranges have been published for different regions and variety-rootstock combinations. Failla et al. (1993a) proposed one common sufficiency range using a dataset which contains data from thirteen varieties grafted on ten different rootstocks, grown at four locations in northern Italy on different soil types. Failla et al. (1993b) proposed reference values for several cultivars and regions in north-central Italy. Ciesielska et al. (2002) in the Piedmont region (Italy) developed individual standard values for leaf nutrients for two cultivars, ‘Barbera’ and ‘Nebbiolo’. Stringari et al. (1997) proposed standards by developing homogeneous groups of grape varieties from different appellations from northern and central Italy. Pacheco et al. (2010) developed specific ranges for ‘Tricandeira’ grafted on Richter 99 for the Portuguese region of Borba (Alentejo). Failla et al. (1995) proposed standards for each of the most important varieties located in eight subzones within the Tuscany region. Considering a long-term survey in representative viticultural regions in Italy, standards have been proposed for macronutrients and trace-elements in leaf blades at fruit-set and veraison (Bavaresco et al., 2010). Finally, García-Escudero et al. (2013) proposed reference values for ‘Tempranillo’ in the Rioja AOC of Spain. In spite of the regional character of all these references, they are widely used for nutritional diagnosis around the world, even for different varieties.

The accuracy of the ranges obtained improves when the number of foliar analysis in the dataset increases and when variation sources such as variety and rootstock combinations, seasonal weather conditions, soil types and cultural practices (Robinson, 2005) decrease. These sources of variation in nutritional status often generate wide reference ranges (Sumner, 1978). Therefore, the most accurate reference ranges are achieved at a local scale (Failla et al., 1995; Robinson, 2005). Furthermore, many authors even proposed a yearly adjustment of the references to consider the impact of seasonal weather conditions over the years (Failla et al., 1993a).

‘Red Grenache’ is one of the most common wine grape varieties in the world, and the second in the Rioja region. This is an area of Spain with a strong wine tradition, well known for producing top quality wines. The aim of this study was to develop reference levels for the nutritional diagnosis of ten essential nutrients for leaf blades and petioles of ‘red Grenache’ grapevine grafted on Richter 110, at the flowering and veraison phenological stages. Those general reference levels will allow us to improve the accuracy of the ‘red Grenache’ nutritional diagnosis in order to design better fertilization programs. Also, considering that the proposed ‘red Grenache’ references were calculated from populations with optimal yield and grape juice quality parameters, the use of these references could contribute to the improved quality of the final wines by means of the general improvement of the nutritional status of the vineyards grown with ‘red Grenache’.

Materials and methods

1. Sampling area and establishment of the data bank

A data bank of blade and petiole analysis was compiled over seventeen years (1992-2008) in the north-eastern area of Spain within the Rioja AOC (1º 40' 55'' to 2º 54' 46'' W – 42º 4' 24'' to 42º 38' 15'' N). A total of thirty-six representative fully productive vineyards (more than 5 years old) of Vitis vinifera L. cv. red Grenache (Garnacha tinta, Grenache noir) grafted on Richter 110 were selected to create a representative data bank in order to obtain reference levels for the whole region. The data bank included different descriptive parameters of each vineyard: (i) plot characteristics (year of planting, geographic coordinates, altitude, percent slope, chemical soil analysis), (ii) vineyard design (spacing, planting density, row orientation, training system), (iii) mineral nutrient concentration in leaf tissues (blade or petiole) and sampling time (flowering or veraison), (iv) plant material (variety, rootstock), (v) cultural practices (vineyard training, fertilization, hydric regime, soil management), (vi) yield, and (vii) environmental factors (pests, diseases, climatic incidents and physiological alterations). The dataset included data from five environmental subzones and the different soils found within the Rioja AOC. In all, the ‘red Grenache’ dataset consisted of a total of 950 blade and petiole analyses.

The vineyards chosen for this study had mainly non-irrigated and mechanically-tilled soils. Planting density ranged between 2,700 and 3,205 vines·ha^-1, and the training systems were Gobelet, Vertical Shoot Position (VSP) Double Cordon Royat, and VSP Guyot. In general, most of the vineyards had optimal production (higher than 3,000 kg·ha^-1) and the average bunch weight was 236 g. Grape quality was within the usual values for the Rioja AOC. Average total soluble solid content was 22.5 ºBrix, pH was 3.36, and titratable acidity was 6.87 g·L^-1 expressed as tartaric acid. The average cumulative annual rainfall was 451 mm for the period 1992-2008.

2. Soil properties

Soil textures ranged between loam, sandy loam and clay loam (Soil Survey Staff, 2010), and chemical analysis showed a generally low organic matter concentration (Walkley-Black method) that ranged between 0.5 and 1.7 g·100g^-1 dry weight (d.w.). Cation exchange capacity (C.E.C.) (extraction by sodium acetate 1 M and Na determination by flame emission spectrometry) was within acceptable limits (5.95 and 19.3 mmol_c·100g^-1 d.w.), as well as active limestone, determined by the Drouineau method (0.20 and 12.6 g·100g^-1 d.w.), and total carbonates (Bernard calcimeter method), which ranged between 2.05 and 40.6 g·100g^-1 d.w. The pH (1:5, 25ºC) was high, between 7.90 and 8.65, while electrical conductivity (E.C.) (1:5, 25ºC) was low, between 0.08 and 0.40 mmhos·cm^-1. Exchangeable bases K, Ca and Mg in soil (1 M ammonium acetate extraction) were determined by flame atomic absorption (Ca and Mg) and flame emission (K) spectrometry. The concentration of bases for the soils included in the dataset was, in general, adequate for grapevine cultivation: Ca ranged between 8.2 and 22.2 mmol_c·100g^-1 d.w., Mg ranged between 0.32 and 1.53 mmol_c·100g^-1 d.w., and K ranged between 0.62 and 5.91 mmol_c·100g^-1 d.w. On the other hand, P (Olsen method) was higher than 5.38 mg·kg^-1 d.w.

3. Leaf sampling

A homogeneous subplot of 450 vines was selected from each vineyard and leaf blades and petioles were collected twice per growing season, at flowering and veraison. Thirty leaves from different sunlight exposure were randomly collected within each subplot. One leaf per plant was taken from a fruit-bearing shoot of average vigour. At flowering, leaves opposite to the first bunch were chosen and at veraison leaves opposite to the second bunch were selected due to an early aging of basal leaves. Therefore, the leaf opposite to the first bunch could be inadequate for plant nutritional evaluation at the beginning of veraison (Romero et al., 2010).

4. Sample preparation and mineral analyses

Leaf blades and petioles were separated, washed three times with tap water, rinsed with distilled water, oven-dried (Dry-Big; J.P. Selecta, Barcelona, Spain) at 70ºC for 48 h, ground with an ultracentrifugal mill (ZM1; Retsch, Haan, Germany) to pass through a 0.50-mm mesh screen, and finally stored at room temperature to be analyzed. For chemical nutrient analysis, 0.200 g of the ground sample were used.

Nitrogen (N-organic + N-NH₄⁺) was analyzed by the Kjeldahl method (Horneck and Miller, 1998) after mineralization in 5 mL H₂SO₄ with a catalyst (K₂SO₄ + CuSO₄·5H₂O + 2% Se) mixture at 370ºC for 45 min. Subsequently, NH₃ was distilled, collected in 2% H₃BO₃ and titrated with HCl 0.025 N. For chemical analysis of the other nine nutrients, dry samples were wet-digested with 3 mL 95% H₂SO₄ and 4 mL 30% H₂O₂ by the microwave method (Hoenig et al., 1998). Phosphorus, K, Ca, Mg, Fe, Mn, Zn, Cu and B were determined by Inductively Coupled Plasma-Atomic Emission Spectrometry (Optima 3000DV; PerkinElmer, Norwalk, CT, USA). Double deionized water (Milli-Q; Millipore, Bedford, MA) was used for all dilutions. Concentrations were expressed in terms of dry weight, using g·100g^-1 for macronutrients and mg·kg^-1 for micronutrients.

5. Statistical data analysis

Prior to the analysis, the dataset was checked to eliminate anomalous data due to unhealthy vines, young vineyards less than six years old or outliers (higher or lower than ± 3σ from the average value). Therefore, data from vineyards that could affect the variability due to their age or their sanitary status were ruled out. However, yield and must quality were not used as discriminative factors to eliminate data before the calculation of both SR and DOP references. This was based on the assumption that the objective was to obtain general references for the ‘red Grenache’ variety and the selected vineyards had, in general, optimal yields for this region.

Data were statistically evaluated from a descriptive statistical approach (modified from Stringari et al., 1997; García-Escudero et al., 2013). The procedure began with the verification of the normal distribution for each nutrient by means of the Kolmogorov-Smirnov non-parametric test, as a prior step to study the distribution of the population as a whole, with respect to the average value and standard deviation.

With respect to the SR method, the reference ranges that characterize the different nutritional status of the dataset were delimited by means of μ ± k·σ, where the constant k is calculated for each percentage in normal distributions where the average is 0 and the variance 1. Population was divided into five subgroups, considering the central 20% population (μ ± 0.25σ) as the optimal reference level for each nutrient and 60% (μ ± 0.84σ) of the central population to show the populations with higher and lower nutrient contents with respect to the optimal range (García-Escudero et al., 2013). Furthermore, DOP references were obtained from the central value (μ) of the dataset selected for each nutrient.

When a normal distribution was not verified for a specific nutrient, data was transformed logarithmically and the SR subgroups and the DOP reference for each nutrient recalculated from the average value of the Log dataset.

Finally, when the Log-transformation to normal distribution did not normalize the distribution, the percentiles (P20, P40, P60 and P80) were calculated to obtain the SR subgroups while the DOP reference value which represented the optimal status for each nutrient was calculated using the median of the population, or percentile P50.

Data analysis was performed using SPSS (version 15.0; SPSS Inc., Chicago, IL, USA).

Results

1. Comparison of the ‘red Grenache’ dataset with respect to international references

Tables 1 and 2 show the percentage of samples from the ‘red Grenache’ leaf blade dataset that are within the optimal range according to references published by other authors for different tissues, phenological stages, grapevine cultivars and regions; Tables 3 and 4 show the same analysis for petiole.

For leaf blade at flowering (Table 1), the dataset optimal percentage for N ranged from 7.8% with respect to the Trentino references (Italy) (Failla et al., 1993a) to 98.8% of the optimal values when using the Bordeaux references (Loué, 1990). Phosphorus ranged from 20.4 to 98.2%, K from 19.8 to 100%, Ca from 21.6 to 92.8%, Mg from 13.2 to 97%, Fe from 23.0 to 100%, Mn from 12.0 to 99.4%, Zn from 0.6 to 18.7%, Cu from 0 to 93.8%, and B from 2.5 to 97.5% (Table 1).

Leaf blade at veraison (Table 2) and leaf petiole at flowering (Table 3) and veraison (Table 4) also showed strong differences between diagnosis results using international references on the ‘red Grenache’ dataset.

For leaf blade at veraison (Table 2), the dataset optimal percentage ranged between 18.5 and 98.4% for N, 27.1 and 100% for P, 7.4 and 93.1% for K, 22.8 and 98.4% for Ca, 18.5 and 100% for Mg, 13.3 and 100% for Fe, 23.0 and 100% for Mn, 11.6 and 58.1% for Zn, 1.2 and 100% for Cu, and 11.6 and 87.8% for B. Therefore, the biggest difference between ranges was observed for Cu between the Italian references (Bavaresco et al., 2010) and the Trentino references (Failla et al., 1993a).

With respect to petiole at flowering, N ranged from 6.6 to 97%, P from 0 to 100%, K from 26.3 to 100%, Ca from 9.0 to 99.4%, Mg from 15.0 to 100%, Fe from 4.4 to 95.6%, Mn from 5.2 to 70.7%, Zn from 0 to 59.3%, Cu from 20.8 to 98.1%, and B from 1.7 to 100%. In this case, the biggest difference between dataset optimal percentages was observed for P using ‘Cabernet sauvignon’ references from Brazil (Fráguas et al., 2003) and Bordeaux references (Loué, 1990).

For petiole at veraison, N ranged from 2.6 to 99.5%, P from 17.5 to 96.3%, K from 12.2 to 96.8%, Ca from 3.7 to 100%, Mg from 13.8 to 100%, Fe from 10.2 to 99.2%, Mn from 13.3 to 65.6%, Zn from 4.7 to 51.6%, Cu from 8.7 to 71.4%, and B from 17.2 to 100%. The largest difference between dataset optimal percentages was observed for N using ‘Tempranillo’ references (García-Escudero et al., 2013) and Bordeaux references (Loué, 1990). Considering the Bordeaux references (Loué, 1990), 99.5% of the ‘red Grenache’ dataset would be in an optimal nutritional status for N, while using the ‘Tempranillo’ references 97.4% would be in a non optimal nutritional status, in spite of the fact that ‘Tempranillo’ references were suggested for the same region as the ‘red Grenache’ dataset.

2. Calculation of SR references

Nitrogen in both tissues and at both phenological stages, K, Ca and Mn in leaf blade at flowering and veraison, B in leaf blade at flowering, P, Ca and B in petiole at flowering, and Zn and B in petiole at veraison showed a normal distribution.

The Log-transformation to normal distribution was effective for P, Fe and Cu in leaf blade at flowering, Fe and B in leaf blade at veraison, P and Fe in petiole at veraison, and K, Fe, Mn, Zn and Cu in petiole at flowering.

Finally, P20, P40, P60 and P80 were calculated in cases where the Log-transformation to normal distribution was unlikely. That was the case for Mg and Zn in leaf blade at flowering and P, Mg, Zn and Cu in leaf blade at veraison. For leaf petiole, percentiles were calculated for Mg at flowering as well as for K, Ca, Mg, Cu and Mn at veraison. The calculated percentiles were used to define the 20% and 60% of the central data.

The sufficiency ranges for ‘red Grenache’ grapevine calculated for ten essential elements in leaf blade and petiole at flowering and veraison are shown in Tables 5 to 8.

3. Calculation of DOP references

The procedure to obtain the reference ranges was the same as the SR references. For DOP references, the reference value for each nutrient was the population mean or the mean value from the Log-transformation of the data to normal distribution. When the Log-transformation to normal distribution was ineffective, the DOP reference value was calculated using the median of the population, or percentile P50.

Table 9 shows the DOP reference values for ‘red Grenache’ macro and micronutrients in leaf blade and petiole at both flowering and veraison. The coefficient of variation (CV(%)) for each reference DOP index is also shown.

The CV(%) was higher for petiole than for blade for N, P, K, Mn and Zn at both phenological stages, Mg at flowering, and Ca and Fe at veraison. However, Ca showed similar CV(%) at flowering for both tissues. On the other hand, Cu and B at both phenological stages, Fe at flowering, and Mg at veraison showed higher CV(%) for blade than for petiole.

4. Comparison of ‘red Grenache’ references with respect to international references

A detailed comparison between the optimal ranges (central 20% population) obtained for ‘red Grenache’ in Rioja with respect to the different international references found the following:

Leaf blade at flowering

‘Red Grenache’ references for leaf blade at flowering (Tables 1 and 5) showed higher values in the following cases: N and P when compared to Trentino (Failla et al., 1993a) and general Italian references (Bavaresco et al., 2010), and B when compared to references for the Trentino region and north-central Italy (Failla et al., 1993a, 1993b).

On the other hand, ‘red Grenache’ references showed lower values for P and Mg when compared to references for Australia (Weir and Cresswell, 1993), K when compared to Australia (Weir and Cresswell, 1993) and the Trentino region (Failla et al., 1993a), Ca when compared to vines from north-central Italy (Failla et al., 1993b), Zn when compared to Trentino, north-central Italy, Australian and Italian references (Failla et al., 1993a, 1993b; Weir and Cresswell, 1993; Bavaresco et al., 2010), and Cu when compared to north-central Italian, Australian and general Italian references (Failla et al., 1993b; Weir and Cresswell, 1993; Bavaresco et al., 2010).

With respect to the sensitivity of the references, Italian optimal ranges for all the nutrients studied had wider ranges than those obtained for ‘red Grenache’ in Rioja (Bavaresco et al., 2010).

In addition, the optimal ranges for N, P, K, Ca, Mg, Fe and Mn for north-central Italy (Failla et al., 1993b), Ca, Mg, Fe, Mn and Cu for the Trentino region (Failla et al., 1993a), the macronutrients for the Bordeaux region (Loué, 1990), as well as for N, P, Ca, Mg, Mn and B for Australia (Weir and Cresswell, 1993) also showed wider values than ‘red Grenache’ optimal ranges. For many of the other nutrients, besides showing wider ranges, their references were shifted compared to those of ‘red Grenache’ (Tables 1 and 5).

Finally, the comparison of ‘red Grenache’ references for blade with ‘Tempranillo’ references in the same winemaking region (García-Escudero et al., 2013) showed higher values for K, Fe, Mn and B and lower values for N, P, Ca, Mg, Zn and Cu (Tables 1 and 5).

Petiole at flowering

With respect to leaf petiole at flowering (Tables 3 and 6), ‘red Grenache’ references showed higher N values than Bordeaux, Australian and Portuguese (‘Tricandeira’) references (Loué, 1990; Robinson et al., 1997; Pacheco et al., 2010), higher Mg values than Portuguese (‘Tricandeira’) and Brazilian (‘Cabernet sauvignon’ and ‘Italian Riesling’) references (Pacheco et al., 2010; Fráguas et al., 2003), and higher Ca, Fe and B values than Portuguese references (Pacheco et al., 2010).

Furthermore, ‘red Grenache’ references showed lower K values than Australian references (Robinson et al., 1997), lower Zn values than Australian, Brazilian and Portuguese references (Robinson et al., 1997; Fráguas et al., 2003; Pacheco et al., 2010), and lower Mn values than Brazilian references (Fráguas et al., 2003).

On the other hand, international references showed wider optimal ranges than those for ‘red Grenache’. In this sense, P, K, Mn and Cu for Portuguese references (Pacheco et al., 2010), N, P, K, Ca, Mg, Fe, Cu and B for Brazilian ‘Cabernet sauvignon’ and ‘Italian Riesling’ references (Fráguas et al., 2003), and all the studied macronutrients for Bordeaux (Loué, 1990) showed wider optimal ranges than ‘red Grenache’ references (Table 6). Furthermore, for some nutrients, like Mg for Brazilian ‘Cabernet sauvignon’ and ‘Italian Riesling’ as well as N for Bordeaux, besides showing wider ranges, their references were shifted compared to those of ‘red Grenache’ (Tables 3 and 6).

Finally, ‘red Grenache’ references presented higher N, P, Ca, Mg, Fe, Mn and B values compared to ‘Tempranillo’ references for the same winemaking region in Spain (García-Escudero et al., 2013). On the other hand, Zn and Cu showed similar optimal ranges, while K showed a wider optimal range than ‘red Grenache’ references (Table 6).

Leaf blade at veraison

‘Red Grenache’ references for leaf blade at veraison (Tables 2 and 7) showed, for the optimal range, higher K, Ca and Cu values than Italian references (Bavaresco et al., 2010) and higher B values than Trentino references (Failla et al., 1993a). On the other hand, the same references showed lower N values than Australian references (Weir and Cresswell, 1993), lower P values than Trentino and Australian references (Failla et al., 1993a; Weir and Cresswell, 1993), lower Zn values than Australian, Italian, and north-central Italian references (Weir and Cresswell, 1993; Failla et al., 1993b; Bavaresco et al., 2010), and, finally, lower Cu values than north-central Italian references (Failla et al., 1993b).

Furthermore, ‘red Grenache’ had narrower optimal ranges than Trentino (Failla et al., 1993a), Italian (Bavaresco et al., 2010), Bordeaux (Loué, 1990), Australian (Weir and Cresswell, 1993) and north-central Italian references (Failla et al., 1993b) for all nutrients except Cu from Italy. Also for leaf blade at veraison, some of the nutrients showed shifted ranges from those obtained for ‘red Grenache’. This was the case for B from Trentino (Failla et al., 1993a), K, Ca and Cu from Italy (Bavaresco et al., 2010), Zn and Cu from north-central Italy (Failla et al., 1993b), and N and Zn from Australia (Weir and Cresswell, 1993).

On the other hand, ‘Tempranillo’ references for the same region (García-Escudero et al., 2013) showed higher values for N, Ca, Mg, Zn and Cu, lower values for K, Fe, Mn and B, and similar P values as ‘red Grenache’ references (Tables 2 and 7). Furthermore, ‘red Grenache’ showed shifted ranges for all nutrients, except for P, from those obtained for ‘Tempranillo’.

Leaf petiole at veraison

The Brazilian ‘Cabernet Sauvignon’ references (Fráguas et al., 2003) showed higher Zn values than ‘red Grenache’ while the ‘Italian Riesling’ references (Fráguas et al., 2003) showed higher Zn and Cu and lower Mg values than ‘red Grenache’ references (Tables 4 and 8). Finally, Italian references showed higher K values and lower Ca, Mg and Cu values than ‘red Grenache’ references (Bavaresco et al., 2010).

On the other hand, ‘red Grenache’ references showed narrower N, P, K, Ca, and Mg optimal ranges than Bordeaux, Italy, and Brazilian references for cvs. ‘Cabernet sauvignon’ and ‘Italian Riesling’ (Loué, 1990; Fráguas et al., 2003; Bavaresco et al., 2010). Furthermore, narrower Fe, Mn, Zn, Cu and B optimal ranges than those from Brazilian and Italian references were obtained, with the exception of Cu from Italy. Furthermore, for Zn from Brazilian ‘Cabernet sauvignon’ and Mg, Zn and Cu from Brazilian ‘Italian Riesling’, besides showing wider ranges, their references were shifted from those of ‘red Grenache’ (Tables 4 and 8).

Finally, ‘red Grenache’ references for leaf petiole at veraison (Tables 4 and 8) showed higher and shifted values for N, P, K, Ca, Mg, Fe, Mn and Cu and slightly higher and shifted values for Zn and B than ‘Tempranillo’ references (García-Escudero et al., 2013).

Discussion

1. Comparison of the ‘red Grenache’ dataset with respect to international references

It should be considered that all the surveyed grapevines included in the dataset were of good productive, sanitary and vegetative status. Thus, low diagnosis coincidences are incongruent with respect to a dataset that corresponds to representative ‘red Grenache’ vineyards.

The comparison shows important differences. For example, the biggest difference in the interpretation of the nutrient status of the ‘red Grenache’ dataset, using blade at flowering, was observed for B with respect to references from Trentino (Failla et al., 1993a) and Australia (Weir and Cresswell, 1993). According to these references, the ‘red Grenache’ dataset would be in an optimal nutritional status for B when using references given for Australia (Weir and Cresswell, 1993), while its status would be non optimal when compared to Trentino references (Failla et al., 1993a) (Table 1).

The high variability found for the dataset comparisons indicates the importance of using adequate references to determine the nutritional status of a determined cultivar. Thus, references must be adapted to each variety, soil and climatic condition, even within the same winemaking region. This fact has also been highlighted by other authors (Failla et al., 1993a, 1995; Ciesielska et al., 2002).

In general, the Bordeaux references (Loué, 1990) covered a greater number of optimal cases for all nutrients, while the Spanish references for ‘Tempranillo’ (García-Escudero et al., 2013) covered fewer cases (Tables 1-4). This is due to a broader optimal range of nutrient concentration in Bordeaux references, while ‘Tempranillo’ references have more limited and perhaps sensitive optimal ranges, specific to a variety with an earlier maturation and senescence than ‘red Grenache’.

2. SR and DOP reference concentrations

The database was split to assure that SR and DOP references would be specific to the combination Vitis vinifera L. cv. red Grenache grafted on Richter 110 and to reflect the total production and agro-climatic conditions of the region where references had been obtained.

The SR method divided the data population into five classes (low, below optimal, optimal, above optimal and high) considering the central range as the optimal nutritional status, whereas the higher and lower ranges of nutrient content with respect to the optimal range suggest that a corrective fertilization program should be considered.

However, the use of Cu-based products in vineyards for phytosanitary purposes prompts the adsorption processes of Cu by the leaf surface. This adsorption increased the total Cu concentration analyzed and therefore a Cu reference with a real physiological meaning for DOP as well as for SR was not obtained. This fact is critical at veraison as by then the phytosanitary applications have already started. The CV(%) for Cu's DOP references shows the high variability found for both tissues at veraison (Table 9).

Each nutrient reference for the DOP method is the mean or centred value, which represents the dataset population for each nutrient as a whole (Table 9). Concentrations below or above the DOP reference value, for each nutrient, produce negative or positive DOP indexes, respectively, and therefore a corrective fertilization program or a reduction of the nutrient in the fertilization programs must be considered.

The lower CV(%) of DOP references for blade suggests that blade has a higher sensitivity than petiole to show deficiencies or excesses of N, P, K, Mn, and Zn at both flowering and veraison stages, Mg at flowering, and Ca and Fe at veraison. On the other hand, petiole has higher sensitivity than blade to show Cu and B deficiencies or excesses at both phenological stages, as well as to show deficiencies of Fe at flowering and Mg at veraison.

Therefore, a technician can analyze leaf blade or petiole, or analyze both tissues and evaluate each nutrient in the most appropriate tissue for a more accurate nutritional diagnosis of ‘red Grenache’ according to the corresponding reference, DOP or SR in this case, for the phenological stage of the sampling.

In addition, SR or DOP references will be less reliable when the regional conditions of the studied vineyard are different from the ones where references were originally obtained (Failla et al., 1995; Robinson, 2005; García-Escudero et al., 2013) or if references are employed for the nutritional diagnosis of other varieties. The proposed references will be less accurate when the vineyard conditions differ from the ones selected to create the dataset.

3. Comparison of ‘red Grenache’ references with respect to international references

The sensitivity of the references could be estimated by means of the width of the optimal ranges. In summary, the comparison (detailed in the Results section) of international references (Tables 1-4) with respect to ‘red Grenache’ SR references from Rioja (Tables 5-8) showed, in general, wider optimal ranges for international references for both tissues and phenological stages. Furthermore, important differences were observed between ‘red Grenache’ DOP references (Table 9) with respect to the nutrient concentration values proposed by those international references (Tables 1-4). Both considerations mean that the use of non-appropriate references has a lower sensitivity to detect deficiencies or excesses in ‘red Grenache’. Furthermore, Spanish references for ‘Tempranillo’ from the same winemaking region also showed important differences with respect to ‘red Grenache’ references, reinforcing the need to establish specific references for the different winemaking regions and even between different cultivars within a geographic area.

Conclusion

SR and DOP references are proposed for the combination Vitis vinifera L. cv. red Grenache grafted on Richter 110 in order to reflect the nutritional status of vineyards under similar production and agro-climatic conditions as the region where references have been obtained. The proposed references will be a useful tool to assess the nutritional status of ‘red Grenache’ around the world, but they will be less accurate when the rootstock, the variety or the regional conditions of the studied vineyard differ from the ones selected to create the dataset; therefore, a proper use of the references requires a technician to evaluate the differences regarding soil, climate, rootstock, as well as other factors.

Finally, references for ‘red Grenache’ will be enhanced in the future because of a continuous increase of the foliar database as well as the introduction of new criteria in the calculation, such as must quality and/or different yield ranges. In this sense, a bigger database would make it possible to obtain more specific references, such as references for rainfed and irrigated vineyards or references that take into account different training systems, soil managements and the age of the vines.

Acknowledgements: This study was supported by the National Institute of Agricultural Research (Spain) and the Regional Government of La Rioja (Spain) with Projects INIA-SC00-016, PR-01-03, PR-01-04 and PR-03-05, among others. We also thank Mrs. M. Carmen Arroyo, the staff at the Regional Laboratory of La Grajera (La Rioja), the staff at the Viticulture and Oenology Section (SIDTA-ICVV), and the vine growers who helped develop the database.

References

• Bavaresco L., Gatti M. and Fregoni M., 2010. Nutritional deficiencies, pp. 165-191. In: Methodologies and Results in Grapevine Research. Delrot S. et al. (eds.), Springer, Heidelberg (Germany). doi:10.1007/978-90-481-9283-0_12
• Champagnol F., 1990. Rajeunir le diagnostic foliaire. Prog. Agric. Vitic., 107, 343-351.
• Ciesielska J., Malusa E. and Bourlot G., 2002. Grape leaf diagnosis in the Piedmont Region. Acta Hort., 594, 387-392. doi:10.17660/ActaHortic.2002.594.49
• Failla O., Stringari G., Porro D. and Scienza A., 1993a. Determination of leaf standards for apple trees and grapevines in northern Italy, pp. 37-41. In: Optimization of Plant Nutrition. Fragoso M.A.C. and Beusichem M.L. (eds.), Kluwer Academic Publishers, Dordrecht (Netherlands). doi:10.1007/978-94-017-2496-8_7
• Failla O., Stringari G., Porro D. and Scienza A., 1993b. Stato nutrizionale di alcune zone viticole dell’Italia centro-settentrionale. VigneVini, 3, 77-82.
• Failla O., Scienza A., Stringari G., Porro D., Tarducci S. and Di Dionisio A., 1995. Indagine sullo stato nutrizionale dei vigneti in alcune zone toscane a denominazione d’origine per una proposta d’interpretazione delle analisi fogliari e del terreno. VigneVini Suplemento, 12, 15-25.
• Fráguas J.C., Miele A. and Silva E.B., 2003. Grapevine nutritional diagnosis methods for the “Serra Gaucha” viticultural region, Brazil. J. Int. Sci. Vigne Vin, 37, 15-21.
• García-Escudero E., Romero I., Benito A., Domínguez N. and Martín I., 2013. Reference levels for leaf nutrient diagnosis of cv. Tempranillo grapevine in the Rioja Appellation. Commun. Soil Sci. Plant Anal., 44, 645-654. doi:10.1080/00103624.2013.745385
• Hoenig M., Baeten H., Vanhentenrijk S., Vassileva E. and Quevauviller P., 1998. Critical discussion on the need for an efficient mineralization procedure for the analysis of plant material by atomic spectrometric methods. Anal. Chim. Acta, 358, 85-94. doi:10.1016/S0003-2670(97)00594-1
• Horneck D.A. and Miller R.O., 1998. Determination of total nitrogen in plant tissue, pp. 75-84. In: Handbook of Reference Methods for Plant Analysis. Kalra Y.P. (ed.), CRC Press, Boca Raton, FL (USA).
• Kliewer W.M., 1991. Methods for determining the nitrogen status of vineyards, pp. 133-147. In: Proceedings of the International Symposium on Nitrogen in Grapes and Wine, 18-19 June, Seattle, WA (USA).
• Kodur S., 2011. Effects of juice pH and potassium on juice and wine quality, and regulation of potassium in grapevines through rootstocks (Vitis): a short review. Vitis, 50, 1-6.
• Loué A., 1990. Le diagnostic foliaire (ou pétiolaire) dans les enquêtes de nutrition minérale des vignes. Prog. Agric. Vitic., 107, 439-453.
• Lucena J.J., 1997. Methods of diagnosis of mineral nutrition of plants. A critical review. Acta Hort., 448, 179-192. doi:10.17660/ActaHortic.1997.448.28
• Monge E., Montañés L., Val J. and Sanz M., 1995. A comparative study of the DOP and the DRIS methods, for evaluating the nutritional status of peach trees. Acta Hort., 383, 191-199. doi:10.17660/ActaHortic.1995.383.19
• Montañés L., Heras L., Abadia J. and Sanz M., 1993. Plant analysis interpretation based on a new index: deviation from optimum percentage (DOP). J. Plant Nutr., 16, 1289-1308. doi:10.1080/01904169309364613
• Pacheco C., Jordao P.V., Vieira S., Santos F., Comenda J., Santos M.C., Roque Vale C. and Prates M.A., 2010. Preliminary reference values for leaf analyses of Vitis vinifera Trincadeira/99R in the Portuguese region of Borba/Alentejo. Acta Hort., 868, 225-230. doi:10.17660/ActaHortic.2010.868.28
• Robinson J.B., Treeby M.T. and Stephenson R.A., 1997. Fruits, vines and nuts, pp. 347-389. In: Plant Analysis - An Interpretation Manual. Reuter D.J. and Robinson J.B. (eds.), Inkata press (Australia).
• Robinson J.B., 2005. Critical plant tissue values and application of nutritional standards for practical use in vineyards, pp. 61-68. In: Proceedings of the Soil Environment and Vine Mineral Nutrition Symposium, 29-30 June 2004, San Diego, CA (USA). Christensen L.P. and Smart D.R. (eds.), ASEV, Davis, CA (USA).
• Romero I., García-Escudero E. and Martín I., 2010. Effects of leaf position on blade and petiole mineral nutrient concentration of Tempranillo grapevine (Vitis vinifera L.). Am. J. Enol. Vitic., 61, 544-550. doi:10.5344/ajev.2010.09091
• Soil Survey Staff, 2010. Keys to Soil Taxonomy (11th Edition). USDA-Natural Resources Conservation Service, Washington DC (USA).
• Stringari G., Porro D., Failla O. and Möller F., 1997. A method to assess and manage leaf analysis standards according to genetic and environmental variability. Acta Hort., 448, 145-154. doi:10.17660/ActaHortic.1997.448.22
• Sumner M.E., 1978. Interpretation of foliar analysis for diagnostic purposes. Agron. J., 71, 343-348. doi:10.2134/agronj1979.00021962007100020028x
• Weir R.G. and Cresswell G.C., 1993. Plant Nutrient Disorders 1: Temperate and Subtropical Fruit and Nut Crops. Inkata press (Australia).