**3. Results**

**Table 1** and **Figure 1** show that EHUs with high yield and high weight of pruning wood (vigorous EHUs) correspond with low probable alcohol grade, low polyphenolic content, and high acidity. These results were to be expected according to numerous previous studies by other authors [19, 33].

The AHC analysis has allowed grouping the 28 analyzed variables of the surface horizon (results not shown) into 3 homogeneous classes, represented by 1 of its variables. Thus, the fine land (FL); the alpha index (AI), which indicates the exchange capacity of the clay; and the humidity at field capacity (H33C) represent, among other variables, the silt content, the total limestone, and the pH; the content in sand, in clay, and in organic matter and the electrical conductivity and the cation exchange capacity; and the content of coarse elements, the active limestone, and the moisture content of the wilting point, respectively.

Observing the biplot graphics of the PCA (**Figure 1**) carried out for the EHUs (observations) and the characteristics of the grape studied together with the three representative variables of the epipedon, it is possible to differentiate between three groups of EHUs. The first group is formed by C1, C2, D2, and D3, which is characterized by its vigor, high yields, and higher levels of malic acid; the second group is composed of B1, B2, and D1 that could be considered as transitional; and the third group is represented by A1, A2, and A3 with a higher probable alcohol level, anthocyanin content, and total polyphenol index (TPI). Regarding the three elements of

**181**

**Table 1.**

**Figure 1.**

the H33c and malic acid (**Figure 1**).

**3.1 Between-vineyard scale**

*Terroir Zoning: Influence on Grapevine Response (Vitis vinifera L.) at Within-vineyard…*

**UHM Yield WP %vol pH M A TPI ANT** A1 835 abc 230 a 14.45 bc 3.95 c 2.3 abc 54.5 ef 706 cde A2 1082 cd 407 cd 14.4 bc 3.86 ab 1.96 a 56.5 f 734 de A3 706 a 231 a 14.76 c 4.09 d 2.92 cd 52.5 f 788 e B1 1022 bcd 390 c 13.68 a 3.87 ab 2.01 ab 46.8 bcd 612 abc B2 767 ab 308 b 13.86 ab 3.82 ab 2.17 ab 48.6 cd 655 bcd C1 2342 g 559 e 13.35 a 3.9 bc 2.97 cd 43.6 abc 562 ab C2 2439 g 584 e 13.40 a 3.87 ab 2.7 bcd 40.6 a 530 a D1 1275 de 307 b 13.45 a 3.8 a 2.3 abc 49.9 de 673 cd D2 1640 f 690 f 13.84 ab 3.81 a 3.09 d 43.0 ab 597 abc D3 1506 ef 461 d 14.02 ab 3.83 ab 2.6 bcd 46.1 bcd 637 bcd Sig. \*\*\* \*\*\* \*\*\* \*\*\* \*\* \*\*\* \*\*\*

*Results (mean 2012–2015) of each UHM: yield (g), weight of pruning wood (WP, g), probable volumetric alcohol degree (%vol), pH, malic acid (MA, g/l), total polyphenol index, and anthocyanins (ANT, mg/l).*

the soil epipedon and its associated variables according to the AHC, it was found that there are positive correlations between the FL and the content in malic acid and between the H33c and the pH of the grape, as well as negative correlation between

*Biplot graphs of observations (EHU) and variables related to the vineyard (yield, WP,%vol, pH, MA, TPI, and ANT) and to the epipedon (FL, AI, H33C). F1, F2, and F3 are the first three factors of the PCA.*

Depending on the scale used in the class discrimination of the NDVI, different maps can be obtained, although the values of the distribution are constant. Thus, **Figure 2** presents the graphic representations of the four vineyards with a common classification of values, in contrast to the FIA drawn at the plot scale. On the contrary, in **Figure 3** a classification of the individualized NDVI values has been carried out for each vineyard. Comparing both figures it is observed that in **Figure 3** the NDVI distributions in each vineyard follow a pattern more similar to the FIA; in addition there are greater contrasts, something that facilitates the zoning (**Figures 2A** and **3A**). In particular, vineyards A and D present a spatial distribution

*DOI: http://dx.doi.org/10.5772/intechopen.86444*

*Terroir Zoning: Influence on Grapevine Response (Vitis vinifera L.) at Within-vineyard… DOI: http://dx.doi.org/10.5772/intechopen.86444*


### **Table 1.**

*Plant Communities and Their Environment*

States Department of Agriculture has been used [30].

the very low class and the fifth quantile to the very high class.

describing the profile and analyzing the different horizons in the laboratory. In this way, between two and three pits per hectare of vineyard are made, density suitable for very detailed soil zoning [12, 32]. The soil classification proposed by the United

The NDVI is defined as the difference between the radiance value in near infrared and red, divided by their sum [20]. In this work, these radiance values have been obtained from multispectral images captured by the Pléiades satellite (0.5 meters of spatial resolution) on August 25, 2014, and August 19, 2015, and by the SPOT 5 satellite (2.5 meters of spatial resolution) on August 14, 2013. The calculation and graphic representation of the NDVI are carried out pixel by pixel, with the help of the ArcGIS 10.1 software from the Environmental Systems Research Institute (ESRI). The definition of the classes (very low, low, medium, high, and very high) is done according to five quantiles, the first quantile corresponding to

The statistical analysis of the data was carried out through principal component analysis (PCA) and univariate ANOVA, after checking normality and homogeneity of variances of the variables. The significance of these analyses was determined for the probability levels p < 0.05 (\*), p < 0.01 (\*\*), and p < 0.001 (\*\*\*). The means are compared by the Duncan test when there were significant differences in the analysis of variance. The SPSS program, version 15.0 (SPSS Inc. Chicago, Illinois), was used for the ANOVA analyses, and for the rest of the statistical calculations, the XLSTAT 2019.1.2 supplement was used on Microsoft Excel 2007. This complement was also used to perform an agglomerative hierarchical clustering (AHC) reducing the 28 analyzed variables of the epipedon before performing the PCA. In this case, the correlation between variables has been calculated for a significance level alpha = 0.05. For the geostatistical study of the NDVI distribution, the normalized Moran index (NMI) is used, which measures the spatial autocorrelation allowing to evaluate if the NDVI pattern is clustered, dispersed, or random. For the calculation of this index, as well as the associated z-value, the ArcGIS 10.1 ESRI tool is used.

**Table 1** and **Figure 1** show that EHUs with high yield and high weight of pruning wood (vigorous EHUs) correspond with low probable alcohol grade, low polyphenolic content, and high acidity. These results were to be expected according

The AHC analysis has allowed grouping the 28 analyzed variables of the surface horizon (results not shown) into 3 homogeneous classes, represented by 1 of its variables. Thus, the fine land (FL); the alpha index (AI), which indicates the exchange capacity of the clay; and the humidity at field capacity (H33C) represent, among other variables, the silt content, the total limestone, and the pH; the content in sand, in clay, and in organic matter and the electrical conductivity and the cation exchange capacity; and the content of coarse elements, the active limestone, and the

Observing the biplot graphics of the PCA (**Figure 1**) carried out for the EHUs (observations) and the characteristics of the grape studied together with the three representative variables of the epipedon, it is possible to differentiate between three groups of EHUs. The first group is formed by C1, C2, D2, and D3, which is characterized by its vigor, high yields, and higher levels of malic acid; the second group is composed of B1, B2, and D1 that could be considered as transitional; and the third group is represented by A1, A2, and A3 with a higher probable alcohol level, anthocyanin content, and total polyphenol index (TPI). Regarding the three elements of

to numerous previous studies by other authors [19, 33].

moisture content of the wilting point, respectively.

**180**

**3. Results**

*Results (mean 2012–2015) of each UHM: yield (g), weight of pruning wood (WP, g), probable volumetric alcohol degree (%vol), pH, malic acid (MA, g/l), total polyphenol index, and anthocyanins (ANT, mg/l).*

### **Figure 1.**

*Biplot graphs of observations (EHU) and variables related to the vineyard (yield, WP,%vol, pH, MA, TPI, and ANT) and to the epipedon (FL, AI, H33C). F1, F2, and F3 are the first three factors of the PCA.*

the soil epipedon and its associated variables according to the AHC, it was found that there are positive correlations between the FL and the content in malic acid and between the H33c and the pH of the grape, as well as negative correlation between the H33c and malic acid (**Figure 1**).

### **3.1 Between-vineyard scale**

Depending on the scale used in the class discrimination of the NDVI, different maps can be obtained, although the values of the distribution are constant. Thus, **Figure 2** presents the graphic representations of the four vineyards with a common classification of values, in contrast to the FIA drawn at the plot scale. On the contrary, in **Figure 3** a classification of the individualized NDVI values has been carried out for each vineyard. Comparing both figures it is observed that in **Figure 3** the NDVI distributions in each vineyard follow a pattern more similar to the FIA; in addition there are greater contrasts, something that facilitates the zoning (**Figures 2A** and **3A**). In particular, vineyards A and D present a spatial distribution

### **Figure 2.**

*Spatial distribution of NDVI from Pleiades image. A common classification is carried out for the four vineyards. The labels (A, B, C and D) represent the vineyards of the same name.*

### **Figure 3.**

*Spatial distribution of NDVI from Pleiades image. A classification of the individualized NDVI values has been carried out for each vineyard. The labels (A, B, C and D) represent the vineyards of the same name.*

of the NDVI similar to the FIA (**Figure 3**). However, plots B and C seem to have a more chaotic NDVI distribution in relation to the FIA, something that seems to be related to the modification of the terrain before planting. This is also the reason why

**183**

in three EHUs:

medium classes appear mainly.

*Terroir Zoning: Influence on Grapevine Response (Vitis vinifera L.) at Within-vineyard…*

making the classification of the vineyard scale distribution (**Figure 3A**).

vigorous, in relation to the rest of the UHMs in the vineyard.

For the study of intra-vineyard variability, we will focus on plot D (**Figure 4**). The difference between the characteristic profiles of each EHU can be observed, appreciating, at first sight, the fluventic character of the D2B profile (**Figure 4B**) and typical of areas of deposition of material or fertile ground. A unique feature of this type of profile is that its agricultural behavior is similar to an addition of slow liberalization fertilizer, which will influence increasing soil fertility and vigor and vineyard production (**Table 1**). Specifically, this profile has a content in organic matter in the epipedon (Ap) of 2.25%, while in the Ab horizon (52–80 cm deep), there is a level of 3.6%. Regarding the NDVI, sector D2 associated with this profile (**Figure 3D**) is included mainly in the high and very high classes. This indicates, as well as the vegetative, grape composition, and pedological results, that it is a UHM that can be characterized as

In **Figure 4D** a NDVI classification has been carried out independently in each EHU; the result obtained is not technically operational, at least in a vineyard of the size of the studied one (2 ha). In order to refine the zoning delimitation to this scale of work, we could redefine the EHUs with the help of the NDVI distribution from the classification of the vineyard (**Figure 4E**). In this way vineyard D will be zoned

D1: northeast facing slope whose characteristic profile is a mesic Calcic Haploxeralf with 80 cm effective depth and with an argillic horizon (20–80 cm of depth). It is the least productive EHU of the vineyard, whose characteristics of the grape are high probable alcoholic degree, lower acidity, and high content in total polyphenols and anthocyanins (**Table 1**). Regarding the NDVI, very low, low, and

D2: material deposition area characterized by fine loamy, mixed, mesic, and Typic Xerofluvent with 145 cm effective depth and darker color of the epipedon. It is expected that it is the freshest sector (and most sensitive to frosts) in relation to the other two sectors of the vineyard. All these factors mean that production is the highest of the three sectors and that the grape composition is the one with the lowest

plot C, which according to **Table 1** includes the most vigorous UHMs in the study, does not include more surface of the high or very high classes of the NDVI map

Considering **Figure 2**, most of vineyard B is included, according to the NDVI, in the very high class, while the results of production and composition of the grape (**Table 1**) do not correspond to this result, agreement with previous studies [16, 19, 34]. Also sector D2 (and part of D3) is included in the very high class (**Figure 2**), but in this case this EHU is characterized by having a high yield and a high pruning weight, as well as low sugar levels, pH, IPT and anthocyanins, and the highest content in malic acid; this coincides with the aforementioned works. Again, it seems that the modification of the natural characteristics of the environment prevents the use of the NDVI as a tool for the zoning of the vineyard, at least on this scale of work. On the contrary, sectors A1 and A3 of vineyard A are included in the very low class (**Figure 2A**). In comparison with the other sectors (**Table 1**), the vines of A1 and A3 have shown the lowest weight of pruning wood, low yields, high levels of probable alcoholic degree, pH, IPT, and total anthocyanins, as well as low levels of malic acid. The other sector of plot A (A2) has been discriminated against by the NDVI since it is not included in the very low class. Sector A2 (**Table 1**) presents higher values of production and weight of pruning wood than sectors A1 and A3. In any case, this discrimination of EHUs is better defined (in relation to the FIA) when

*DOI: http://dx.doi.org/10.5772/intechopen.86444*

(**Figure 3**).

**3.2 Intra-vineyard scale**

*Terroir Zoning: Influence on Grapevine Response (Vitis vinifera L.) at Within-vineyard… DOI: http://dx.doi.org/10.5772/intechopen.86444*

plot C, which according to **Table 1** includes the most vigorous UHMs in the study, does not include more surface of the high or very high classes of the NDVI map (**Figure 3**).

Considering **Figure 2**, most of vineyard B is included, according to the NDVI, in the very high class, while the results of production and composition of the grape (**Table 1**) do not correspond to this result, agreement with previous studies [16, 19, 34]. Also sector D2 (and part of D3) is included in the very high class (**Figure 2**), but in this case this EHU is characterized by having a high yield and a high pruning weight, as well as low sugar levels, pH, IPT and anthocyanins, and the highest content in malic acid; this coincides with the aforementioned works. Again, it seems that the modification of the natural characteristics of the environment prevents the use of the NDVI as a tool for the zoning of the vineyard, at least on this scale of work.

On the contrary, sectors A1 and A3 of vineyard A are included in the very low class (**Figure 2A**). In comparison with the other sectors (**Table 1**), the vines of A1 and A3 have shown the lowest weight of pruning wood, low yields, high levels of probable alcoholic degree, pH, IPT, and total anthocyanins, as well as low levels of malic acid. The other sector of plot A (A2) has been discriminated against by the NDVI since it is not included in the very low class. Sector A2 (**Table 1**) presents higher values of production and weight of pruning wood than sectors A1 and A3. In any case, this discrimination of EHUs is better defined (in relation to the FIA) when making the classification of the vineyard scale distribution (**Figure 3A**).

### **3.2 Intra-vineyard scale**

*Plant Communities and Their Environment*

**182**

**Figure 3.**

**Figure 2.**

of the NDVI similar to the FIA (**Figure 3**). However, plots B and C seem to have a more chaotic NDVI distribution in relation to the FIA, something that seems to be related to the modification of the terrain before planting. This is also the reason why

*carried out for each vineyard. The labels (A, B, C and D) represent the vineyards of the same name.*

*Spatial distribution of NDVI from Pleiades image. A classification of the individualized NDVI values has been* 

*Spatial distribution of NDVI from Pleiades image. A common classification is carried out for the four vineyards.*

*The labels (A, B, C and D) represent the vineyards of the same name.*

For the study of intra-vineyard variability, we will focus on plot D (**Figure 4**). The difference between the characteristic profiles of each EHU can be observed, appreciating, at first sight, the fluventic character of the D2B profile (**Figure 4B**) and typical of areas of deposition of material or fertile ground. A unique feature of this type of profile is that its agricultural behavior is similar to an addition of slow liberalization fertilizer, which will influence increasing soil fertility and vigor and vineyard production (**Table 1**). Specifically, this profile has a content in organic matter in the epipedon (Ap) of 2.25%, while in the Ab horizon (52–80 cm deep), there is a level of 3.6%. Regarding the NDVI, sector D2 associated with this profile (**Figure 3D**) is included mainly in the high and very high classes. This indicates, as well as the vegetative, grape composition, and pedological results, that it is a UHM that can be characterized as vigorous, in relation to the rest of the UHMs in the vineyard.

In **Figure 4D** a NDVI classification has been carried out independently in each EHU; the result obtained is not technically operational, at least in a vineyard of the size of the studied one (2 ha). In order to refine the zoning delimitation to this scale of work, we could redefine the EHUs with the help of the NDVI distribution from the classification of the vineyard (**Figure 4E**). In this way vineyard D will be zoned in three EHUs:

D1: northeast facing slope whose characteristic profile is a mesic Calcic Haploxeralf with 80 cm effective depth and with an argillic horizon (20–80 cm of depth). It is the least productive EHU of the vineyard, whose characteristics of the grape are high probable alcoholic degree, lower acidity, and high content in total polyphenols and anthocyanins (**Table 1**). Regarding the NDVI, very low, low, and medium classes appear mainly.

D2: material deposition area characterized by fine loamy, mixed, mesic, and Typic Xerofluvent with 145 cm effective depth and darker color of the epipedon. It is expected that it is the freshest sector (and most sensitive to frosts) in relation to the other two sectors of the vineyard. All these factors mean that production is the highest of the three sectors and that the grape composition is the one with the lowest

### **Figure 4.**

*Characteristic profiles of the EHUs: 4A (sector D1), 4B (sector D2), and 4C (sector D3). NDVI classification carried out independently in each EHU and location of the pits (4D). Definitive* terroir *zoning redefined with the NDVI, with a single classification for the three EHUs (4E).*

### **Figure 5.**

*Spatial distribution of the NDVI in the years 2014 (satellite Pleiades) and 2013 (satellite SPOT 5) and proposed terroir zoning.*

probable alcoholic degree, the highest acidity, and the lowest content of total polyphenols and anthocyanins (**Table 1**). The very high and high classes of the NDVI are the most present in this EHU.

D3: hillside with southwest orientation, whose main soil is fine loamy, mixed, mesic, and Typic Xerothent, with an effective depth of 145 cm. It could be considered as the EHU of intermediate characteristics between the other two.

**Figure 5** shows the graphic representation of the NDVI from other years (2014 and 2013), verifying the temporal stability of the zoning carried out in vineyard

**185**

pixel.

possible.

particularly in the harvest.

*Terroir Zoning: Influence on Grapevine Response (Vitis vinifera L.) at Within-vineyard…*

D. It is noteworthy that the image of the year 2013 is of a resolution (2.5 meters) lower than the rest of the years and the distribution pattern of the NDVI is adjusted to the proposed zoning. In order to achieve a harvest as homogeneous as possible, we could recommend practices aimed at reducing vigor in D2 and increasing it in

The NMI calculated for the NDVI distribution has a value of 0.998 and a z-score of 397 for the years 2015 and 2014, while in 2013 the value of the NMI is 0.999 and the z-score is 77. For 3 years there is less than 1% likelihood that this clustered pattern could be a result of random chance [35]. In this type of spatial analysis, it seems that the spatial resolution of the starting image does influence, with a lower correlation with lower spatial resolution. In any case, the NDVI distributions have a grouped pattern indicating that there is a link between the NDVI distribution and the landscape or the distribution of the environment

It has been verified that there is similarity in the distribution of the NDVI and the FIA, provided that both cartographies are made at the same scale. In vineyards grown on man-modified soils, it seems that the use of vegetation indexes, such as the NDVI, does not give the expected results, being able to conclude that there is a link between the NDVI and the characteristics of the environment, in particular

In the within-vineyard *terroir* zoning, the EHU associated to a material deposition area and characterized by the main series fine-loamy, mixed, mesic, and Typic Xerofluvent is related to the very high class of the NDVI and at the same time with vigorous properties of the vineyard: high yield, high weight of pruning wood, low probable alcohol grade, high acidity, and low levels of IPT and anthocyanin content. These results have been obtained in comparison with two EHUs associated with hillside and whose main soil series are mesic, Calcic Haploxeralf and fine-loamy,

Regarding the NDVI, the interannual stability in the pattern has been demonstrated regardless of the resolution of the image, at least from 0.5 to 2.5 m/

Epipedon characteristics related to agronomic results have been found. Thus, it was found that there are positive correlations between the FL (and associated variables such as silt content, total limestone, and pH) and content in malic acid and between H33c (and the associated variables such as the content in coarse elements, the active limestone, and the moisture content of the wilting point) and the pH of the grape, as well as the negative correlation between the H33c and malic acid. In the future, it will be interesting to find a methodology that allows to integrate the analytical results, not only of the superficial horizon but also of all the horizons of the soil profile and the vegeto-productive results and grape composition, to be able to relate them to each other as exhaustively as

The importance of the size of the vineyards to find PV applications is noteworthy. In the case of DOCa Rioja, 87% [36] of the vineyards are small (between 0.1 and 2 hectares), making the sectorization of the vineyards technically and economically unviable in order to carry out localized treatments. In any case, it is advisable to project plantations that, as far as possible, facilitate individualized management,

*DOI: http://dx.doi.org/10.5772/intechopen.86444*

with those related to the soil and landscape.

mixed, mesic Typic Xerorthent.

elements in space.

**4. Conclusions**

D1, always with the limitations of practical management.

*Terroir Zoning: Influence on Grapevine Response (Vitis vinifera L.) at Within-vineyard… DOI: http://dx.doi.org/10.5772/intechopen.86444*

D. It is noteworthy that the image of the year 2013 is of a resolution (2.5 meters) lower than the rest of the years and the distribution pattern of the NDVI is adjusted to the proposed zoning. In order to achieve a harvest as homogeneous as possible, we could recommend practices aimed at reducing vigor in D2 and increasing it in D1, always with the limitations of practical management.

The NMI calculated for the NDVI distribution has a value of 0.998 and a z-score of 397 for the years 2015 and 2014, while in 2013 the value of the NMI is 0.999 and the z-score is 77. For 3 years there is less than 1% likelihood that this clustered pattern could be a result of random chance [35]. In this type of spatial analysis, it seems that the spatial resolution of the starting image does influence, with a lower correlation with lower spatial resolution. In any case, the NDVI distributions have a grouped pattern indicating that there is a link between the NDVI distribution and the landscape or the distribution of the environment elements in space.
